CA2211910A1 - Fingerprint-acquisition apparatus for access control; personal weapon and other systems controlled thereby - Google Patents

Abstract

At a first end of an optic-fiber prism assembly are fiber terminations to contact a relieved surface, e.g. finger (stabilized by a handgrip). In a region where fiber diameter is essentially constant with longitudinal position, light enters the prism, crosses the fibers and enters individual fibers through their sidewalls, lighting the terminations. To allow crosslighting of the assembly, the fiber-optic numerical aperture (NA) is small: preferably not exceeding one-half. Due to fingerprint etc. detail, fractions of light pass along the fibers; at the assembly second end a detector responds with an electrical-signal array based on the surface relief.
The signals are processed to check finger etc. identity and applied to control access to a personal weapon, other equipment, facilities, data, or a money service. FTIR ("frustrated total internal reflection") bright- and dark-field versions have various benefits. For use of small, low-cost detectors - and/or internal-mirror versions that light the finger straight-on - the assembly has a separate element, e.g. a <u>high</u>-NA fiber-optic taper with extramural absorption or "EMA" material (no entry light crosses it). For a weapon, a unitary antibypassing module (which matches a weapon port, and must be present) holds part of the access-control <u>and</u> firing systems. Bullets etc. fire only on a special signal from the module.

Description

W 096124283 PCTrUS96101615 ~PRINT-ACOUISITION APPARATUS FOR A~-C.S ~ ~OL;
PERSONAL WEAPON AND OTHER SYSTEMS CONTROLLED ~ :~Y

BA(.;KWtOIJNI~

1. FIELD OF THE lNV~N'~10~

This invention relates generally to automatic acquisition of fingerprints and oth~r relieved-surface images, for ~c~ec control -and to systems whose ~c~ is controlled by such automatic image acquisition. The invention relates more particularly to fiber-optic prism systems for such print acquisition, and to ~oop~rating m~h~ni-cal and electrical provisions fo- both ~nh~n~; n~ the identity confir-mation and deterring ci~..~ention of the identity confirmation.
Systems to which access is controlled in accordance with the present invention inc:lude personal weapons, other apparatus, facili-ties, fi n~n~j al services and information services.

2. PRIOR ART

Hand w~ro~C - Several U. S. patents address the topic of 25 automatic control of hand w~pon~ to deter unauthorized use. U. S.
4,467,545 of Shaw suggests applying automatic fingerprint analysis.
U. S. 4,003,152 to Barker incidentally proposes use of voiceprints and brainwaves.
Shaw's presentation is ~o~c~ptual, actuallv only outlining many tasks to be ~co~rlished. The hand-weapon environment imposes for-midable constraints of both time consumption and certainty in iden-tification - as well as difficult seCo~Arv considerations including size, weight, shock resistance, pnysical reliabilitv, and cost.

Bodv data - Apar' from the hand-weapon context, hundreds of U. S. patents address the topic of automatic verification of identity through automated acquisition and analysis of body data or voice-prints. Applying these to a hand-weapon envi~o- - t ~m~n~ ~ in ComDination and under adverse field conditions, both (a) nezr-instantaneous accessibility to input data and (b) a reasonably high level of identification cer~ainty.

W096/24283 PCTnUS96101615 Nonoptical fingerprint t~hnologies if econc c may be usable with certain aspects of the preser.t invention. Such t~hni ques appear in U. S. 4,353,056 to Tsikos of Siemens, for a direct capac-itive fingerprint sensor; 4,394,773 to Ruell, also of Siemens, for a 5 piezoelectric fingerprint sensor; 4,526,043 to Boie of AT&T Bell Lahs for a ~p~citive system with fingerprint-force ..od~lated carrier;
4,577,345 to Abramov for an integrated-circuit sensor array overlaid with a pressure-sensitive membrane; and 4,788,593 to OVshi ne~y for a thin-film photosensor array.
More-corventional optical t~hniques - CCD arrays, video etc.
- heretofore have been the most widely analyzed and explored.

Ideal FTIR - These include relatively sophisticated systems that set out to collect information that is - at least in conc~rt -binary (i. e., not gray-scale or multivalued), through application of the principle of frustrated total int~rn~l reflection (FTIR).
There are two main groups of such systems. One is typified by U. S. Patent 4,728,186 to Eguchi of Fujitsu. Eguchi's FTIR-analogue system uses flat lighting, and evidently does not rely on inci-20 dent /detected angle relations, but inherently distingni sh~.cfingerprint ridges from fingerprint y-OOv~S - based upon the inabil-ity of light from grooves to exit toward a detection direction that a critical angle.
More-conventional FTIR systems are described in, for example, 25 U. S. Patents 3,947,128 to W~inh~ger; 4,783,l67 to Schiller; 5,233,-404 to Lolyh~; and 5,210,588 to Lee of Goldstar (Korea). Light is incident on a solid/air interface from within the solid medium at an angle to the interface normal (perpendicular) that is just slightly greater than a critical angle for total internal reflection for the 30 given solid at an interface with air.
The critical angle is arcsin (l/n), where _ is the refractive index of the solid material. In principle all transmission through the interface is prevented; the incident light is all reflected back into the solid.
If some other medium instead of air, however, is juxt~pos~, the effective critical angle rises to arcsin (_'/n), where n' represents the refractive index of the other medium. Now total internal reflec-tion occurs only if the angle of incidence ~cee~ this hiqher critical value. If not, then only some of the light is sp~llarly 40 reflected int~rn~lly, and the r~--in~r passes through.
Conventional FTIR fingerprint systems take advantage of these relationchips by directing incident light to such an interface, from within a solid clear prism or plate, at an angle which is in~
ate between the critical angles arcsin ~l/n) for air and arcsin (n'/_) for typical biological materials such as skin or flesh, and water.
If no finger is pres~ent, or a light ray strikes the surface at z 5 y~GOve of a fingerprint, then the light under consideration is all internally reflected as be~ore.
If instead a li.ght ray strikes the surface at a ridge of a fingerprint, then part of this light passes through the surface and into the material or the finger, where it is scattered diffusely.
A significant fraction (pernaps as much as half of the scattered light) is returned through the boundary surface into the solid clear prism or plate, but ]propagating into a wide directional range - a full h~mi eph~re. Thus only a small fraction of this backscattered light coincidentally travels along the same specular-reflection 15 direction that would be taken, according to total internal reflec-tion, in the ~hS~nce of the finger or at a fingerprint groove.
For practical c:ollection geometry, in theory less than one per-cent of total intensity of light striking a ridge area is collected;
this theoretical low value is subject to important degradations.
In one operating mode, light collected at the sp~7~ internal-reflection ~osition but in the backscattering case can be distin-guished, based upon its lower intensity, from that directed toward the same collector and position in the total-internal-reflection case. This distinction provides one way of distinguishing ridges 25 from y~oov~s, respectively.
Ideally the int:ensity ratio is on the order of a hundred, and is slightly further ~nh:~nc~ by the fact that some light entering a ridge is absorbed. ~-uch a ratio, for this particular way of distin-guishing ridges from y OGv~S, iS high enough to fairly characterize the d-stinction as co~rtually binarv.
In another operating mode, another way to distinguish ridges from grooves is to e:~clude collection of any specularly reflected light. The amount o e backscattered light that can be collected in directions o~her than the internal-reflection direction, although 35 weak, is much greate~e than its h~kground.
In such "other" directions, in idealized principle no light at ~ all is collected in a total-internal-reflection case.
Hence a ~on~ptually binary distinction can be obt~in~ for viewing of the h~kec~ttered light too, at viewing angles away from ~o the sp~c~ r-reflect:ion direction. It is accordingly well known that conceptually binary data can be co;ilected from an FTIR system, using either the so-called:

W096/24283 PCTrUS96/01615 ~ "bright field" case, the first operating mode ~i scllcs~ above, in which air-filled fingerprint yLO~v~S provide (through specu-lar total-internal reflection) 2 bright ~ J-v~lld or "field" on which to view the dark lines corr~cpon~i~g to ridges; or ~ "dark fieldn case, the second operating mode dis~llqs~ above, in which air-filled fingerprint grooves provide (through absence of backscattering) a dark field on which to view the lighter lines corresponding to ridges.

In the two cases light is incident on the finger-contact surface from essentially the same direction; and the interaction between the light, the solid clear prism or plate, and the finger surface relief are essentially the same. What differs is primarily the detector 15 pl~ nt and preferably, in a conventional system, the collection-cone angle (related to numerical aperture).
As to placement, in the bright-field case light is collected for the detector along the direction of sp~c~lAr total-internal reflec-tion. In the dark-field case light is collected in some other 20 direction - most c~ ~nly, but not n~cess~rily, about the normal to the solid/air interface surface.

Practical FTIR - Neither the supposedly continuous contact of a fingerprint ridge ~g~inct the glass surface nor the theoretically clean separation of a fingerprint y~OOv~ from the glass is perfect.
Due to these imperfections, observed intensity - and therefore contrast and siqnal-to-noise ratio - can depart dramatically from predictions of the simple geometrical theory intro~lc~ above. These departures in turn lead to significant preferences as between dark-30 and bright-field operation.
FTIR contrast - On a microscopic level, fingerprint ridges have myriad tiny gaps due to fibrous nature of the skin and presence of sweat glands or pores. Such gaps create tiny pockets of FTIR specu-lar reflection where simple theory predicts only scattering. Some incident light that should be scattered is instead specularly re-flected. Resulting bright spots along a ridge cause the average or apparent intensity of the ridge - viewed from the bri~ht-field col-lection position - to rise toward that of the bright groove field.
Resulting intensity at a ridge region, in bright-field opera-tion, is typically far greater than the predicted one percent of the incident light: in adverse cases it can be as high as seventy-five percent.

W096/24283 PCTrUS96101615 _5_ Contrast, often defined in terms of m~ m and minimllm in-tensities I~ and In~n as a ratio C = (I~-I ~n) / ~I~+I ~n), can then fall as low as one-seventh (0.14).
When viewed as backscatter from a dark-field position, the 5 fibrous and pore-in~l~c~ ridge gaps create a series of dark spots along the ridges - uhich on average darkens the ridges from their ideal medium level of brightness. If there is some ~ ~ound level (that is, ii.- the dar]c field is not perfectly dark) this dark~ni n~
lowers tne effective contrast relative to the dark grooves - but at least in the dark-field case if care is taken to avoid significant ~g~ound the contrast will be good, whereas contrast in the bright-field case is poor intrinsically. Contrast is at least potentially better in the dark-f:ield system.
Grooves too can, be troublesome, but only where the grooves are 1s unusually shallow or contain a relatively large amount of liquid, dirt or skin detritu~s.
Most often, intensity variations great enough to be troublesome occur in the variably-dark ridge areas of the bright-field case -but not in the relat:ively darker, or more consistently dark, y~oova 20 areas of the dark-field case. Co~c~quently the intensity is ade-~uatelv well control:Led in:

~ qroove areas oi both systems - bright for bright-field systems and dark for dark-field systems; but ~ ridqe areas of dark-field onlY.

Thus dark-field syste~ms are preferred for higher contrast.

FTIR siqnal-to-noise ratio - In addition to this preference based upon considera1_ion of contrast itself, the same preference arises from consider:Lng the signal-to-noise ratio - where the noise is mainly shot noise.
It can be shown that signal-to-noise is proportional to con-trast. In the dark-iield case contrast C is almost always unity and signal-to-noise is nearly always optimum.
~ In the bright-field case, contrast can be as low as about l/7, with proportionately lower signal-to-noise. The only way to c ,---sate for this signal--to-noise deficiency is to increase the exposure (light level or time,. or their product), by about 72 = fortv-nine times. Real FTIR systems therefore require careful ~ngi n~ing to optimize performance.

CA 022ll9l0 l997-07-30 W096/24283 PCTrUS96/01615 FTIR focal svstems - For micr~mini~ture applications, even more troublesome practical limitations intrud~. A primary obstacle to use of FTIR clear prisms or plates in very close quarters is need for a focal element such as a lens to image the FTIR data onto a detector.
A lens requires focal dist~nces, at both sides, ordinarily totalling several times the focal length. Where f is the focal length and M the multiple of ~ni fication or reduction, the total is (2 + M + 1/M)f. For short-focal-length systems _ roughly equals the transverse-diagonal ~i cion of the object. For a representative fingerprint, that ~im~ncion is 1~ to 2 cm and a miniml1m optical-train length is about 7 cm.
To constrain the detector size and cost, the lens can be used to reduce the fingerprint image at the detector, but then the optical train lengthens to very roughly six focal lengths, or over 10 cm.
15 Optical-path folding is awkward in a small space.
A lens also is susceptible to depth-of=field shallowness, and distortion, particularly severe if the lens and object plane are not reasonably parallel and co~ . Just such geometry is typical in bright-field FTIR devices, in which collection must be off-normal by 20 more than the critical angle vs. air.
The top and bottom of the image have different m~gnifications, and are badly out of focus when the center of the image is fo~llc~, unless the aperture is rather small. The state of the art in minia-ture portable apparatus ordinarily dictates mi ni mizing battery size 25 and m~Yimizing battery life by nimizing lamp power - which is to say, m~imizinq light-gathering power and therefore aperture.

FTIR fiber-oPtic svstems - Two prior patents have proposed substitution of a fiber-optic prism for a clear prism as a dark-field 30 FTIR fingerprint collection block in a fingerprint reader: U. S.

4,785,171 and 4,932,776 of Dowling et al. This substitution offers relief in controlling some or all of the lens-system problems dis-cussed above, but suffers from severe vulnerability to scattering by con~min~tion in the '171 svstem and severe geometrical problems in 35 the '776 systems.
The '776 patent uses a fiber-optic taper, inteqra~ with the fiber prism, to match the print image to a relatively small CCD
array. Also a CCD array is attached directly to the end of the fiber taper remote from the finger.
The fiber core has refractive index 1.62 and the ~ ing 1.48, yi~l~ing ~g~in~t air a moderately high numerical aperture NA = 0.66 and critical angle of about 38~. This choice is conventional for W096/24283 p~l~r '01615 ob~ining good light--gathering power, althol~h many skilled arti C~n~
in this field would prefer a considerably higher numerical aperture.
Dowling applies the full capabilities of the tapered fiber prism to shorten the optica~l system, erect the i s ge plane (supplying an 5 image that is merely ~n~m~rphic but in uniform focus and free of major aberration), an,d eliminate or ~inimi2e effects of cont~min~tion and jarring. Unfortunately, his prism is covered by a CCD at one end and a finger at the other, leaving no suitable entry point for illumination.
Dowling proposes three alternative solutions. In a first, the light is applied to the opposite side of the finger to be read -which in Dowling's term is thus "transilluminated". This option appears unsatisfactory: bones in the finger degrade the uniformity of illumination, FTIR binary character is imr~ired or forfeited, and the optical path is longer and more elaborate.
In a s~con~ o~ Dowling's alternatives, small lamps "might be implanted in the face of the image sensor". There readily ~pr~s no way of either (a) protecting the s~neor from fogging by the lamps, or (b) causing light that o-iginates outs-de any fiber to enter into that fiber - in the sense of being ducted along it to reach the finger end of the prism. This s~co~ alternative, not illustrated, would seem to be inoperative.

Dowling's third alternative is to direct the light into the 25 sides of the fiber pr:ism - sp~ifically from the narrower sides of the taper - propagat:ing toward the finger-contacting surface to be illuminated. In part:icular his ill~mina-ion is di-ected into por-tions of the taper wh~are fiber diameter is changing rapidly with respect to longit~ position (i. e., the part of the taper that is actually tapered). Dowling te~h~C that the light should be thus projected into the pr~sm from all four opposed sides (fo-mi ng in effect a sc~uare light source), and at ~3~c to 45c relative to its major longitl~A~ n~l axi.s".
Analysis indicates that this Dowling system will at best work - ~5 very poorly, and most likely not at all. In particul~r! efficiency of light injection in this ~-nn~r is extremely poor.
~ Some very small fraction of light rays may possibly be injected into the inflection point, the most strongly tapered region, of the taper section. ~hrough succ~ss~ve internal reflections in the transition zone (from that angled region to the larger straight s-_ - t o the taper), rays so injected may possibly enter the ducted mode of propagation along the fibers and so reach the finger surface.

_ _ _ _ _ _ _ _ _ _ W096/24283 PCT~S96/01615 Very high illumination from any given narrow directional cone would be required to effectively illuminate the target finger. To accomplish even this, however, it would be necess~-y to use a taper that has no absorbing material outside the individual-fiber walls ("e~l ral absorbing" or "EMA" material). Much of the returning backscatter from fingerprint ridges would correspondingly esca~e from its fibers at that same inflection region of the taper, and badly fog the image.
Any effort to overcome this problem as by, for instance, using 10 fiber of much higher than usual numerical aperture would be coun-terproductive: the injection m~h~ni e~ described above, marginal at best, would be foreclosed.

FinqerPrint analvsis - A very great body of patent and other literature relates to the interpretaiion and particularly _ _-rison of fingerprint data once acquired. These may be performed visually or by automatic processors.
C~mp~risons in most known systems proceed by abstractins minu-tiae from collected im~g~S Minutiae and their relationchips are Zo usually invariant with age, weight change, position, and geometrical distortion. Massive amounts of data handling, however, are n~e~ to in fact sllcr~ssfully abstract identi~ications from a raw fingerprint image. Portions of the present invention are compatible with ~-ki ng identifications through analysis of minutiae.
Finqerprint stabilization - Some prior-art ~ ntS constrain-ing the finger from which a print is to be read, particularly to control the position of the finger in relation to a sensing surface.
Little attention has been devoted to constr~i ni ng the orientation and 30 fi~n~cs of finger application to the sensor. This omission intro-duces considerable uncertainty in identification, even when s~lppos~
topological invariants such as minutiae are to be det~rmi n~, Bvpass control, and microminiaturization - As to personal-35 weapon ~cress generally, the art fails to cope with a thief whobypasses ("hot wires") or ~~ v~s ~cc~s-control devices, m~k- ng a weapon work without them.
Further the prior art has failed to provide an optical finger-print reade- module ~m~hle to microminiaturization for ~cc~ss control in highly ~m~i ~g field applications, particularly includ-ing personal w~pone - and also ~nco~r~sing ~rrQ~s to use of porta-ble c~..~uLers and phones.

CA 022ll9l0 l997-07-30 W096/24283 PCTnUS96/01615 Public phones, phone credit systems, vehicles, automatic tellers and facility-entry ~CC~55 devices would also be --ni ~gfully ~nce~
by provision of a microminiaturized reader.

SUMM~RY OF THE DISCLOSURE
.

l. INTRO~u~ lON

The system described in Dowling '776 is inop~rative, or margin-ally operative, for these four main reasons:

(a) Liqht should be iniected at a ~avorable place - As a general rule, illumination that is initially outside an optical fiber can be systematicall~r made to pass along the fiber - in the ducted-light, wd~yuide ~-n--~r - only by intro~l~;ng the light through an end of the fiber. If light instead passes into a fiber through a side wall, its entry path is by definition not that of a waveguide-ducted ray, and so ill general (that is to say, on average) the major 20 part of this light wiill find a like exit path nearby and leave the o fiber through the side wall.
Only a very marginal, energy-inefficient exception to this general rule is encountered in the case of a fiber-optic taper (or very strongly cur~red pipe). Reliance upon such an inefficient 25 injection m~h~ni sm should be avoided, to prevent the associated energy waste, image i-oggins, signal-to-noise degradation, and h~ph~7-ard results.
Hence if light is to be injected through a side wall to illumi-nate a finger or like object in contact with a fiber t~r~in~tion 30 (end), the light should be injected so tha- ~t p~cs~s through the side wall i ~ telv adiacent to that t~rmin~tion. In other words, to reach the t~rmin~t:ion the light should not rely on p~cs~ge duct-wise alona the fiberr but instead simply should directly strike the t~rmi~tion imm~iatq3ly after p~s~g~ through the side wall -35 preferably with no dt~cting reflections at ~il along the fiber.
The Dowling '776 system departs from this requirement.
Analogously if light is to enter a fiber through 2 side wall forsubsequent reflection, at a reflective t~rmin~tion (end), into the ducting wd~e~uide operation of the fiber, the light should first pass through the side wal3. i -~iatelv adiacent to that reflecting t~rmin~tion. Again, to reach the t~rmin~tion light should not rely on - and preferably should not undergo - p~cs~ along the fiber W096/24283 PCTnUS96101615 ductwise toward the reflecting t~rmin~tion, but instead simply should directly strike that t~ ~tion ~ tely after p~cs~g~ through the side wall. (After that first end-wall reflection, in forms of the present invention the light is generally ducted by multiple side-5 wall reflections along the fiber toward a finger-contacting t~rmi n~-tion at the other end of the same fiber.) (b) Liqht should be iniected at a favorable anqle - Since illumination is to strike each t~rmin~tion i ~ tely upon entry, 10 preferably without ducting reflection along the destination fiber en route, the illumination should first enter each fiber:

~ at the proper angle for FTIR operation at the t~rm;n~tion (or, in some forms of the invention, first the proper angle for reflection at the t~rmin~tion - after which the fiber does duct the light to the finger-contacting surface at the proper angle for FTIR operation); and ~ preferably in an angular relatio~chip with the fiber which is not favorable to direct entry of the rays into a ducting mode.

These two conditions will now be discl~cse~ in this same order.
To reach the fiber t~in~tions (other than those few which are at the entry face), illumination nec~cs~rily crosses the fiber struc-ture of tne prism, and in the process is strongly diffused. Fortu-nately the initial angle of inclination of the rays relative to the fibers tends to be preserved in propagation of the light.
This operation does require, however, injection of light so that~ once within the prism, it is at the proper FTIR angle (or in some geometries a proper relaying-reflection angle). In general, for the reasons explained in the Prior Art section of this ~o_ 1t, the proper FTIR angle as defined from the surface no-~-l is just greater than the critical angle.
Consideration of representative examples will clarify this. As the critical angle (off normal~ i5 typically just below forty de-grees, injection must be at a steep angle of at least about forty degrees to the norm~l. A good choice for injection is about forty-five degrees (off nor~
In bright-field FTIR imaging, the totally reflected light is to 40 be coupled into the ducting, w~eyuide mode of the fibers, which parallel the prism axis. Assum~ng a right-angle-prism suitable for bright-field operation - with the finger-contacting surface also W096/24283 PCTnUS96/01615 11 _ typically at about forty-~ive dey~ees but relative to the Prism axis - the addition of these two forty-five-degree angles leads to a total of roughly a right angle between the injection path and the prism axis.
In the case of a squared-off prism suitable for dark-field FTIR
imaging, with the finger-contacting surface at right angles to the prism axis, the prism axis and the surface no 1 are cor.y-~ent.
~herefore the inject:ion should be at about forty-five degrees to the prism axis; and must be greater than the critical angle (about thirty-eight degrees~. If the injection is at a shallower angle to the prism ~;s, no total internal reflection can occur.
(In the case of a relaying end-wall reflection too, the injec-tion is best at about 90~ to the prism axis.) The second condLition mentioned above (angular relation unfavor-able to ducting) is :important not only h~ e injection into a taper is wastefully ineffi~:ient, but also for the integrity of a system operating in accordance with the first condition. A direct-crosslit system that allows initially ducting illumination too will yield 20 badly garbled result:s, for the following reason.
It is possible for ducted illumination to be at a correct FTIR
angle, as in certain forms of the present invention - provided that the illumination-car~eying fibers are at the correct FTIR angle to the finger-contacting surface. Such operation, however, requires a fiber-to-contact-surface angle that is different from the analogous angle in a system us:ing direct crosslighting to the finger-contacting fiber t~ n~tions.
Therefore in a system which allowed both direct-crosslit FTIR-angle il umination and ducted illumination, the finger-contacting surface would n~C~qs:l~ily be at an incorrect angle for FTIR opera-tion, relative to one or the other illumination component. The forms of the invention thal: employ ducted FTIR input illumination therefore do r.ot use crosslighl:ing that is direct to the finqer-contactinq t~in~tions, although they do employ crosslighting in another way ~ 35 (direct to a partial:Ly reflecting end wall) as will be seen.
The "s~co~ condition" here under discussion ~mplies mainly that injection should not be into a region of the optical element where fibers are tapered -- or, to state this condition positively, injec-tion should be into a region where fiber diameter is constant with 40 respect to longit~ n~l position along the fiber. Ideally, injection should be arranged to avoid even very strongly curved regions.

W096/24283 PCTnUS96101615 The second condition also has an implication as to systems in which tapers or strongly curved light pipes are used (contrary to the mainly implied condition above). In this case, at least the angle of injection should be kept steep enough, relative to all the changing s angles of fibers within the prism, to preclude entry into ducting.
Dowling' s te~hi ng is to achieve initial ducting of the injected light, directly contradictory to the condition stated here.

(c) The fiber prism should be of a favorable material -10 R~c~ e of the multiple reflections and l~nCi ng effects encountered by injected light in going across the grain of the complex multicylindrical structure, inc ng illumination tends to be strong-ly diffusedi a large fraction can even reverse direction and leave the fiber block through its input face.
This effect is adverse as it amounts to a form of attenuation of the FTIR-usable light. It can leave only a tiny fraction of incident optical power to reach the center (and in one-side illuminated sys-tems the far side) of the fiber prism, at least in usable form, and so tends to illuminate the finger-contacting face very n~nllniformly.
In addition this attenuation effect wastes light, and lamp power - and through massive generation of skew rays will illuminate fiber t~rmin~tions at incorrect angles for FTIR operation. This process diverts light away from FTIR operation, into image fogging and light loss from the system.
It is believed that in fiber-optic elements made with almost all materials - -rcially available now, fogging and light loss would be severe. Generally sp~ki ng, fogging would be so severe as to domi-nate the image data and thereby -ender virtually impossible any extraction of reliable information about fingerprints or the like.
Part of the present invention is based upon recognition that this type of attenuation and fogging can be mi nimi zed by choice of fiber-prism material, and in particular material of suitable numeri-cal aperture - and more specifically that the attenuation length for diffusion across this sort of structure varies, roughly, in-35 versely with the fourth ~ower of the numerical aperture.
In accordance with this recogr,ition, FTIR-usable intensity I at depth x within the prism is roughly related to incident intensity and numerical aperture NA as:

40 I - ~ exp t-2x (NA/2navg)4/D~, (Eq. l) W096/24283 PCTnUS9610161S

where D is the periodicity of the riber structure and n a~ the average of the refractive in~ices of fiber cl~i ng and core.
From the theory of fiber optics, the numerical aperture of optical-fiber materials increases with the difference between refrac-S tive indices of the Eiber core and cl~i ng, according to the knownrelation:
(NA)- = (n core) ~ ~ ~--cla~ng) ~ ~Eq . 2 ) and this holds true Eor a prism made of a fused bundle of optical 10 fibers as well as an unfused bundle or an individual fiber The attenuation length (2_a~/NA)4 D /2 establishes the rate at which FTIR-usable intensity falls off with penetration depth, and more specifically eqlaals the distance over which intensity drops to lJe of an initial value (where e is the base of natural or Naperi-15 an logarithms, and l~_ is about three-eighths). Using this very approximate relation, the attenuation length can be estimated at very roughly 2 mm for NA := 0.66, or 21 mm for NA = 0.35.
In view of this extreme sensitivity, it will now be clear that capability to effectively crosslight a prism depends rigorously upon 20 selection o* a mater:ial with suitably low numerical aperture. As a fingerprint-acquisit:ion 45c prism itself should be roughly 15 to 20 mm across, or in an alternative geometry some lO~ mm, the differ-ence between the above-mentioned values (for attenuation to about three-eighths of initially incident intensity) is sign ficant.
Calculations using Eq. l have been used to define preferable low numerical-aperture v~lues, ~ for marginal operation at the bounds of practicali1ty - and also to define more-highly-preferred very low values n~-~ for more-optimum operation. These values will be presented shortly, in the portion of this section which is more 30 formal; and will be discussed in the Detailed Description section.
(E~. l is only very roughly correct. In particular, for reasons detailed later, Eq. :L deviaies from accuracy at penetration depths greater than a very i~ew ~perhaps only about two] attenuation lengths.
Nevertheless Eq. 1 is entirely adequate for comparisons of perfor-35 mance at different m~merical-aperture values, and also for general discussion of consid~arations affecting system perform~n~e.) The desired range of low numerical-aperture values is not the same for all configurations or definable in the abstract, but rather ~p~n~ on hoth the :illu~ination geometry and the prism size. As 4C pointed out above, the numerical aperture controls the ability of the system to inject FTIR-usable light into the p-ism at various depths.

CA 02211910 199i-07-30 W096/24283 ~CTnUS96/01615 In z practical miniaturized portable apparatus for field use, lamp power must be strictly ~i ni mi zed. On the otner hand the inten-sity of the optical signals re~hi ng the detector should not be too small, as they must compete with waste light r~hing the detector, 5 added to ambient scattered light - sources which may collectively be called optical "fog".
If optical signals are too small, they are buried in this "fog"
- or the resulting electrical signals are ~ by associated electrooptical noise, particularly the shot noise discllcs~ earlier.
In any event this small-signal side of the problem cannot n~C~cS~ily be ci~..vented even by increasing lamp power, as the diffusion type of attenuation under discussion not only robs the system of light n~ for the FTIR process but also diverts that light intc image fog~i ng . Any effort to compensate by raising the 5 overall light level may only make the matter worse, by saturating the detector and so precluding all mea~u~.~nts.
In other words, the type of attenuation which arises from diffusion not only i ~ s the signal level but also threatens the underlying operation of the system. These results of ~cessive 20 attenuation can be escaped only by confronting the numerical-aperture requirements described here.
With a favorable steep-angle, wide-face illumination that is opposed or bidirectional, i. e. projected into a rectangular-ended dark-field prism from sources at two opposite sides, as will be seen 25 the different geometry rather coincidentally leads to low-numerical-aperture selection that is the same as in the unidirectionally lit 45~ bright-field prism. Path leng~h~ni ng due to off-a~is angle just compensates for the shorter distance to the midplane; when power to a single lamp in the 45~ prism geometry is shared between two oppos~
3c lamps in the rectangular geometry, the resulting choice of low numerical aperture comes out the same. Nevertheless, as shown in the Detailed Description section of this document, the oppos~ lighting does confer a striking advantage in illumination uniformity.
FTIR-usable light intensity at the prism center is the resultant 35 of the two usable ~ntensities from the two sources - more specifi- 7 cally, two overl~ppe~ intensity distributions, ~xpo~ntially falling from each opposite direction as set forth earlier in Eq. l. Even a prism illuminated from both sides - possible with the squared-off dark-field configuration - requires effective incident-light diffu-40 sion from each side, at least to the midDlane. For this situation the applicable form of Eq. 1 is two overlapping ~xro~ntial func-tions, tailing off in opposite directions from both entry faces:

W096/24283 PCTrUS96/01615 ~ {exp ~-2x ~N~)4/k] + exp [-2 (XF - X) (NA)4/k]}, (Eq. 3) where XF is the dista~nce to the far side of the prism means.
As will be seen from later detailed discussion, effective 6 illumination even in this ~osed-lighting case is infeasible at Dow-ling's indicated num~!rical aperture of NA = O.66: only the tiniest fraction of incident energy can propagate - maint-i ni ng the original input angle or direction - even halfway across the fiber prism.
Almost as important as the overall attenuation value is the 1O strong nonuniformitv of attenuation as between the edges and midplane of the prism - further complicating image interpretation. At NA =
0.66 even for oppos~ lighting such nonlln;~ormity would amount to a factor of about five thousand between the extreme values, or about +99.98% of the ~ n value.
With proper cho:Lce of low numerical aperture, however, effective illunination is very easily provided. The result is asto i~hi n~l y high FTIR-usable intensity at the midplane and uniformity across the full breadth of the prism ~neans. For example, in the opposed-light-ing case such intensity may vary, across the entire prism m~aans, by a 20 factor on the order of only about l.4, i. e. +15~ of a median value.
Actually Dowlinsl's NA choice is in keeping with usual co~C~rnc for l~ght-gathering power and signal-to-noise ra'io - and in partic-~lar ducting ability - in optical systems, l~ ing to ~ibers of high numerical aperture (high index differential). Conventional wisdom in 2s the art thus t~-h~-e away from the necessary conditions of the g~ ~
try of the present invention.
In addition, at least the injection s~ t of the prism should be free of e~L~ l-a'~sor~tion material.

(d) If a ta~er is used, it should be a se~arate element from the prism - '~hereas the prism should be given a low numerical aperture to mi~i~i ze the diffusion-created attenuation, and should be free of EMA naterial, two o~osite considerations apply to the taper.
First is its pri.mary function of ducting light through a path ~ 35 that is tapered, pref~erably tapered rather tightly. This function tests the ability of the taper to constrain light within its wave-guiding bo77nr~7~7~ies~
Light rays p7Csi ng along an individual tapered fiber ~and so along a tapered fiber bundle), from the broad end toward the narrow end, encounte- incre 7!Ci n~ angles of reflection which are more severe than the environment of alternating increasing and decreasing angles W096/2~283 PCTnUS96/0161S

in a moderately ~u~ved but untapered fiber. These changing reflec-tion angles lead io leakage and interfiber crosstalk.
Ability of a fiber (or bundle) to ~i ni mi ze such signal loss is best ~nhance~ by use of a high numerical aperture. To prevent crosstalk, in r~co~nition that even a high NA will not entirely retain the light, those rays that do escape should be stopped by e~L.~..Iral absorption.
The 5~ron~ consideration -allins fo~ use of hign numerical aperture is the effective degrading of optical energy flow, through a 10 r~--ri ng taper, bv the square of the r~ nification. In practical cases this can amount to loss of more than three-auarters of the input energy, as follows.
The capability of a fiber-optic prism or taper to accept light flux is proportional to the solid angle of the effective acceptance 1~ cone of the element, and this solid angle is in turn proportional to the sauare of the half angle of the cone, just outside the glass.
The half angle, expressed in r~ n~, is roughly ~in air) the numeri-cal aperture NA - so tnat energy flow ove-all goes as the sauare NAZ
of the numerical aperture.
A taper of numerical apertur~ NAtap~r = G.6S and reduction M = 2 to 4, however, passes ~ight from its broad s~ rt to its narrow segment only as if its nu.~erical aperture ha~ an effective value of NAt,per = NA~aper/M = O.33 to O.17 respectively. This reduction in effec'ive numerical aperture is inherent in the geometry of a taper.
2~ Although the numerical aperture NA~per is the same at both ends of a taper, rays that enter the taper within its actual NA~per-estab-lished cone but outside its effective NAt~per-established cone will fall outside tne actual N~per-established cona at the narrow, exit end. (Alons the way from entrance to exit ends, these rays escape 30 from the ducting mode and preferably are trapped ~y EM~ material if provided; if not trapped they can leave the taper laterally as ieakage, or in adverse cases fog the image ~t the narrow end.) Energy flow is therefore proportional to tne effective value NAt~pe~2 = t~At~pgr/M) 2 = C ~ 11 to O.C27 respectively. The roughly c~~r~-able flow throuah a straisht fiber-opti- prism Gr numerical aperture NApr sm = Q 5 is NAprism = Q-_2 Within this apprcximation, the raper of a-tual numerical aper-ture NA~,per = O.66 iherefore transmits nearly all the light from the prism of NApri8m = 0.35 if it is a two-times r~l~ri n~ taper ~M = 2~ -but only (NAta~ ~N~pri3mSZ = O.G27~0.~2 = G.22, ~went~-two percent, of l-ght from the prism if i~ is a fo~r-ti~les r ~in~ taper (M = ~).

W096/24283 PCTnUS96/01615 Hence an even hiqher numerical aperture than the value 0.66 used by Dowling would be preferable for the taper. To match a prism of N~ria~ = 0.35, the taper should have numerical aperture of NA~aper =
N~ri~ ~ M = 1.05 if it is a three-times taper (M = 3), or an even 5 higher value NA~per = l . 4 if it is a four-times taper (M = 4).
Since the prism and taper thus have diametrically conflicting design requirements for practice of the present invention, they are best fabricated as separate elements. As will shortly be seen, a dual-element construction is also ~ hle to incorporation in an 10 entirely novel and effective illumination configuration - even if neither of the elements is a taper.

2. MORE--FORMAL DISC:USSION

(a) A FIRST ASPECT of the invention - In preferred embodi-ments of a first of its independent aspects, the present inventi.on is apparatus for acqUirillg surface-relie~ data from a relieved surface such as a finger.
The apparatus includes prism means formed from optical fibers.
(A fused bundle of fibers is much preferred to unfused fibers, as the latter - with their high-index-differential boundaries between glass and air - attenuate crosslighting much more rapidly.) The prism means in turn include a first end and a second end.
2~ AS Will be seen, the phrase "prism meansn is primarily used to ~n~r~cc important en?boAim~nts of the invention in which two or more fiber-optic optical elements in series are include~ in the optical ~e5~mhly, as well as embodiments having only a single fiber-optic optical element.
The ~irst end comprises t~-min~tions of the fibers for contact with the relieved surface. The eecQn~ end comprises opposite termi-nations of the same or correspo~ g fibers.
By ~Correspon~i ng fibers" here is meant fibers of a second element that may be in series, ac mentioned just above. Such a fiber 35 receives i ght from the ~ibers in tne first element.
A ~COrr~-epOn~i ng fiber" typically is only very roughly aligned with any of the fibers in the first element, so that in practice the light from each fiber in the first element may pass into several fibers of the second -- and each fiber of the s con~ element typi-cally receives light from several fibers of the first. These effects somewhat degrade image resolution, but can be made inco~e~uential by W096/24283 PCT~S96/01615 using prism materials in which the fiber spacing is sufficiently finer than the fingerprint ridge sp~i ng, The second end of the prism means is for p~es~ge of light traveling along the fibers from the first end.
This first major aspect of the invention also includes some means for projecting light across the fiber prism, in a region where the prism satisfies at least one phys-cal condition for efficient non~llcting transillumination, for lighting the first-end t~rmi n~-tions. For breadth and generality in ~is~lesing the invention, these 10 means will be called the "light-projecting means" or simply "project-ing means". Several such "physical conditions" for this lighting region are:

~ Fiber diameter is substantially constant with respect to longi-tll~i n~l positicn.

~ Numerical aperture satisfies this m~Xi m~l~ condition:
NA < 2aaVg(D /xF)1/ , (Eq. 4) where a ,vg - average of core and ~ i ng refractive indices in that region of the prism meansi D - periodicity of the fiber structure in that same region; and X F -- iilumination-path distance across the prism means 2s in that region, and the corv~ntional notation (D /x F) 1/4 means the fourth root of the ratio D /x F -~ Numerical aperture does not ~x~ about one-half.
~ More stringently, numerical aperture does not ~ee~ 0.42.

~ For a prism that is illuminated from both sides, the numerical aperture satisfies a modified form of Eq. (4), NA < 2aavg(D /XM) / ~ (Eq. 5) where XM -- illumination-path distance across the prism means to the Prism midPlane, in the same region.

~ Numerical aperture satisfies this condition:

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _19_ NA ~ 2a avg (D /2x F) / ~ (E~. 6) where the variables are defined as in Eq. 4. This is cimi 1~ to Eq. 4 but has an inserted factor of 2 in the ~ tor of the fourth-rooted function.

- ~ Numerical aperture satisfies the condition of Eq. 5 b~t with 2 factor of ~ similarly inserted.

Sisnificance of the values one-half and 0.42, and of ~qs. 4 through ~o 6, will be clear from detailed discussion later in this document.
Use of a bidirectional, o~ opposed, lighting embo~i - t with the condition given in Eq. 5, and with the same overall lamp power as in a unidirectionally lit embo~iim~-nt that satisfies Eq. 4, favorably rr~r~ ~c variation of lighting intensity across the fingerprint.
(Merely as an exam~le, at NA = 0.35 the variation across a 17~ mm prism may fall from a factor of about 2.3, for unidirectional light-ing, to about 1.4 -- which is to say, from about +~0~ to about +15 of a median value. Even under unidirectional lignting, however, using the present invention the intensity variation is within a range that is readily mi nimi zed by other means as will be shown.) Applying Eqs. 4 and 5, approximations to the illumination dis-tribution across the prism means are easily calculated for various illumination geometries, prism widths and numerical-aperture values.
Representative results appear in the Detailed Description section.
Even though t:he projected light crosses the fibers and is "for illuminating" thei:r first-end termin~tions~ in some forms of the invention as will be seen it does not ~ cs~rily illuminate them directly or i ~i:otely upon fiber entry.
A light fract.ion that is dependent (i. e., whose magnitude is r~ Gr~nt~ on cont~oct between the relieved surface and each iilum~-nated first-end tr~nmin~tion is ducted from that t~ in~tion along its fiber. (By "its f:iber" is mean- the fi~er which is t~r~in~ted by the t~ n~tion. ) The present invention enables such p~Cc~ge of light, to and from the finger-contacting end Oc the prism means, o proceed s~ ccfully according to the well-known principles of FTIR intro~inc~ earlier in this document - d~spite use of fiber-optic prism means.
In addition the apparatus includes some means for receiving -40 at the pri~ ns second end - each light fraction from the first end, and in r~cpo~ce formi ng an electrical signal which is character-istic of the surface relief. Such means accordingly have an W096/24283 PCTrUS96/01615 electrooptical character; here too for generality and breadth these means will be called simply the "electrooptical means".

Although the invention provides significant advances relative to 5 the prior art, nevertheless for greatest enjoyment of the benefits of the invention it is preferably practiced in conjunction with certain other features or characteristics which ~nh~nce its benefits. Among these are, to begin with, three main preferred ~mno~im~nts: a bright-field configuration and two dark-field configurations.

(i) A bright-field : ~o~i -t of the first aspect -In the bright-field FTIR apparatus, the projecting means include some means for projecting light to enter individual fibers, through their respective side walls, i -~iately adjacent to their respective 15 t~rmin~tions.
Each t~rmin~tion, at the first end, is oriented, relative to the projected light and also relative to a longitll~i n~l direction of its fiber, so that the light at that t~r~in~tion is:

20 ~ reflected at that t~min~tion into and along its fiber toward the ele~,ov~tical means, if the relieved surface is out of con-tact with that t~r~ tion; and ~ instead in large part - or at least fractionally - scattered by the relieved surface out of the corr~spon~ing fiber, if the relieved surface is in contact with that fiber t~r~in~tion.

These will be r~ognized as the geometrical conditions for bright-field detection.
Within this first : ~.o~i - t it is preferred that the projecting means direct the light into the prism means at an angle, relative to an axis of the prism means, which is greater than twice the critical angle for the fiber cores ~g~inct air. For purposes of all the ~ i -, the critical angle is defined as the ~ini~ l~ angle 35 off-normal for total int~rn~l reflection.
For an element in which the fibers are straight and untapered, the phrase "axis of the prism means" denotes an ~Yis parallel to the ~ ~ longit~ l direction of the fibers. If the element is a taper, ordinarily the fibers are straight along some line - p~rh~r5 40 most usually the centerline - and this line is the pri~ nc axis.
In other cases, those skilled in the art will understand what defini-tion effectuates the purposes of the invention.

W096/24283 PCTnUS96/01615 It is preferred that the critical angle for the fiber cores ~ nct air be rough~y forty degrees, and that the first end of the prism means lies at an i nrl i n~tion angle of roughly forty-five ~eyrees to the prism axis - and that the projecting means direct the 6 light into the prism means at an angle to the axis of approximately ninety degrees.
In addition it is preferred that the prism means include at least two separately fabricated optical elements secured together along a partially rei-lecting interface: a first one of the elements ~o preferably includes cL fiber-optic prism that includes the first end and that is crossed by the light from the projecting means. A second one of the elements preferably com~rises a fiber-optic taper that in-cludes the seco~ encl.
In this last-mentioned case, more specifically it is preferred that the fiber-optic taper include ext-~-"~ al-a~sorption material, and have a numerical aper~ure greater than one-half - but that the prism have a numerical aperture not ~X~Ai ~ one-half, as abo~e.

(ii) One dark-field embo~i -t of the first aspect - Now 20 turning to a s~co~ preferred ~ 'im~nt of the first aspect of the invention, that : -'i t is a dark-field FTIR apparatus in which the projecting means comprise means for projecting light to enter individual fibers, through their respective side walls, i ~~i~tely adjacent to their re~;pective t~rmin~tions at the first end.
In this case, however, each t~ nAtion at the first end is oriented, relative to the projected light and also relative to a longit~ direction of its fiber, so that the light at that t~r~in~tion is:

~ reflected at that t~-~in~tion out of its fiber, if the relieved surface is out of contact with that t~rmin~tion; and ~ fractionally scattered by the relieved surface into and alonq its fiber towar~d the electrooptical means, if the relieved 3~ surface is in contact with the fiber t~rmin~tion.

~ These are the geometrical conditions for dark-field detection, and are the converse of the bright-field conditions described above for the first ~ o~im~nt~
In this seco~ ~embo~im~nt - still referring to the first "aspect" of the invention - as ir. the firs~ : -'i - t it is prefer-able that the critica.l angle be roughly forty degrees. Here, howev-W096/24283 PCT~S96/01615 er, it is also preferred that the first end of the prism means lie at an inclination angle of roughly ninety degrees to the central axis.
It is further preferable that the light from the light-receiving face of the prism means be incident at an angle greater than roughly 5 forty-five degrees to the central axis, but less than the first-end inclination angle. (For effective FTIR operation the angle of incidence relative to the axis should be, as noted earlier, less than the value of the critical angle aqainst skin.) In this second ~ho~i - t it is preferable that the pri ..~ans 10 numerical aperture be small enough that projected light which crosses the entire prism means have at least roughly one tenth the entering intensity, at least in a region adjacent to the first end. Also applicable to this s~con~ embodiment are the preferred relation set forth in Eq. 4, the m~i numerical-aperture value of one-half, and the two-element construction mentioned above Por the first, bright-field, embo~ t.
F.5p~i ally preferable in use of this second : '-'i - t is an opposed illumination system. In this case ~he projecting means comprise at least two generally opposed light sources for projecting light from opposite sides of the prism means.
In this system light from the projecting means enters the prism means at each side with a respective initial intensity, and some of this projected light from each side crosses the entire prism means.
At least where the light crosses the fibers, preferably the 25 prism-means numerical aperture is small enough that projected light which crosses the entire prism means, from each side, has at least one hundredth of the respective initial intensity.
Generally in this dark-field embo~im~nt of the first aspect of the invention it is highly preferable that the prism means comprise a section having extramural-absorption material. Such a section should be placed between (l) a region where the light crosses the fibers and (2) the electrooptical means.
It is further preferable that this EMA-material section have a numerical aperture which is substantially the same as numerical 35 aperture of the region where tne light crosses the fibers. These EMA-section preferences relate to control of 2 form of stray light that is p~~ ~ to this embo~i -nt of the invention, and will be detailed later in this document.

~iii) Another dark-field : oAim~nt of the first aspect - Now a third preferred embo~ t of the first aspect of the inven-tion is a dark-field FTIR apparatus in which the prism means include W096/24283 PCT~US96101615 at least two separately fabricated fiber-optic optical elements secured together along a partially reflecting interface. A first one of these elements inc~ludes a prism that in turn includes the first end, and that is crossed by the light from the projecting meansi and 5 a ceco~ one of the elements includes the second end.
In this ~ t the projecting means comprise means for projecting light into .he prism to enter individual fibers, through their respective side walls, ~ ately adjacent to the partially reflecting interface. This interface is not at the finger-contacting 10 t~rmin~iions, but rat.her at the opposite end of the prism.
The partially reflecting interface does, however, redirect some of the light from thel projecting means to pass along the individual fibers within the prism toward their respective t~rmi~tions at the first end. Upon rr~ch;ng each t~rmin~tion at the first end, this 15 redirected light is reflected by the t~ ~tion out of the correspo~ing fiber, if the relieved surface is out of contact with that t~rmin~tion; and rractionally scattered by the relieved surface back into the first prism and along the corr~epon~ fiber, if the relieved surface is in contact with that fiber t~m;n~tion.
Once again the conditions described are a form of dark-field geometry, but the illumination in this case is incident on the finger-contacting t~min~tions by ducting along and within the fibers, rather than directly ~y crosslighting. The crosslighting 1~ ;nc an important part of this system, however, for initially injecting the illumination into, the volume of the prism for redirec-tion, by the internal reflector, into the fibers.
As to the light which is fractionally scattered from the finger-contacting surface back into and along the correspon~i ng fiber: this liight passes fraction;~llv through the pariialiy reflecting i.nterface 30 and the s~con~ element to the electrooptical means.
This dual-optica~l-element ~oAim~nt using an internal reflec-tor, like the other ~ ts disc~cs~ above, resolves very elegantly the proble~s of the prior art. Nevertheless it is pre-ferably practiced in .~ way that further optimizes its perfo ~n~
It is advant~geous, though not nec~es~ry, that the s~con~ ele-ment include a fiber-optic taper - preferably with EMA material, and preferably with numer:ical aperture greater than one-half - but yet that the ~rism have mlmerical aperture not ~c~ing one-half. Other preferences as to use of a taper, in other embo~imQnts of the first aspect of the invention as well as this : '-~i nt, are taken up in the next subsection.

W096t24283 PCT~S96/01615 (iv) General preferences as to the first aspect of the invention - All three embo~i -ts described above can reliably and efficiently provide high-signal-to-noise and essentially binary FTIR
data, and thereby resolve important problems l~n~ essed by the prior 5 art. Noneth~l~cs it is preferred that the invention be practiced in conjunction with still other features and characteristics which ~nh~nc~ its benefits and confer still other advantages:

~ Tapered fibers - It is advant~geol~c that the second end of the prism means be different in cross-section from the first end, and that the fibers be tapered to different cross-sections at the second end from that at the first end. By means of this arrangement the first end of the prism means is sized for contact with the surface (the finger or other subject relieved surface~, and the second end is sized for contact with the electrooptical means.
This preferred system is particularly helpful where the cost of electrooptical means - such as, for example, detector arrays -increases steeply with area. For this case the second end of the prism means is smaller in cross-section than the first, and the 20 fibers are tapered to smaller cross-sections at the second end than at the first. Thereby the prism means ~ gnify the surface-relief scattering pattern for application to the electrooptical means. The side face has approximately the same, larger width as the cross-section of the first end of the prism means.
The fiber tapering in such a system ordinarily is localized to one (at least) s~ t of the prism means. In accordance with the general condition - stated above for the first aspect of the inven-tion - that the light be injected at a point where fiber diameter is constant with longit~lAin~ position, the projecting means here direct light into a s~m~nt or the prism means other than that one s~r~nt.
When design is not sensitive to electrooptical --nc cost, the space and weight whicn are required by a taper can be saved; all three above-described embo~im~nts of the invention are ~ ~le to this arrangement. In this regard, the internal-reflector form of the invention is not limited to use with tapers, but rather can be practiced with an additional short straiqht fiber-prism s~m~nt on the electrooptical -nc side of the reflector.

~ Low numerical aperture - Tn the general llni~;~ectional-lighting case it is also preferable that the pri~ ns numericalaperture, at least in a region between the end face and side face, be small enough that projected light which crosses the entire prism W096/24283 PCTnUS96/01615 means have at least roughly one tenth of the initial intensity.
Alternatively the ca,ndition of Eq. 4 may be applied. These relation-ships may be expecte!d to define the limit or margin of advisable or sound operation; the!y correspond to injection of illumination to a 6 depth of roughly twc! attenuation lengths.
For operation lnore solidly within a reliable and easily engi-neered m~nn~r, with ample illumination but better-constrained lamp power, preferably thLe pri~ n5 numerical aperture is small enough that projected light which crosses the entire prism means have at 10 least roughly three-eighths of the initial intensity. Alternatively the numerical aperture satisfies the condition of Eq. (6) above.
Both alternative staLtements, appearing in this paragraph, of prefera-ble relationships correspond to irjection of illumination to a depth of roughly one attenuation length.

~ Finger stabilization - In addition it is beneficial for full enjoyment of the ad~antages of the invention to include some means for stabilizing the human subject's f-nger ~g~inct the first end of the prism means. Preferably these stabilizing means inclu~e a 20 hAn~-ip for firm grasping by the human subject's hand - in such a way that this firm grasping braces a particular finger ~g~inct the first end of the pri.sm means.
For best stabilization it is preferred, ~c~q~ of the m~h~ni CS
of the human hand, t:hat the particular finger be the subject's thumb.
25 Of course c~ifferent stabilizing geometry can be used to ~cc~ te other fingers when .Ippropriate - as for example if the prospective user has no thumb.
Preferably the stabilizing means include some means for orient-ing the finger on the first end of the prism means within a rela-tively narrow range of finger positions. For this purpose it ispreferred that the h~n~grip include some means for orienting the hand, in relation tc~ the first end of the prism means, within a rela-tively narrow range of hand positions.
By virtue of this arrangement, orientation of the hand by the ~ 35 ~ grip _nherently constrains the orientation of the ringer, on the first end of the prLsm means, within the relatively narrow range of finger positions. 'rhis orientation of the hand and finger in turn also inherently tencis to constrain each user to exercise a relatively consistent degree of pressure or firm~ss in applying the finger (thumb) to the prism surface.
Even a modest constraint of finger orientation and fi rm~ec in this fashion can produce important advances in data-processing time CA 022ll9l0 l99i-07-30 W096/24283 PCTnUS96/01615 or certainty of identity confirmation, or both. Thus the invention resolves ambiguities left ~ ressed by Shaw as to fingerprint control of w~po~c in particular, and not adequately addressed heretofore in the fingerprint field generally.
S

~ Processing of received light - Other important groups of preferences in use of the above-described invention relate to use of the light that is collected at the second end of the prism means. As mentioned earlier, the present invention is ~m~n~hle to use with the 10 electrooptical means comprising an optical-detector array, l~i ng to electron,c processing of the image acquired as an array of pixels.
In fact the invention represents a major advance in such detec-tor-array systems, since it permits affixing of the array directly to the second end of the prism means sl~cc~esfully - that is to say, 15 with full realization of near-binary FTIR perfo-~-n~e, and without degradation of optical efficiency or signal-to-noise relations such as seen in the prior ari.
Thus in a preferred form of the invention the electrooptical means include an electrooptical detector array. The array receives, 20 at the second end of the prism means, the light fractionally directed into the fibers at the first end.
In response the detector array provides a corr~spo~ g array of electrical signais, which are characteristic of the surface relief.
In this case the previously mentioned "electrical signal which is characteristic of the surface reliefn ~nS p-eC~e the array of elec-trical signals intro~lç~ here.

~ Processing of resulting electrical signals - In this case preferably the electrooptical means include some means for ~ _~rison 30 of the electrical-signal array with data for a particular specified relieved surface - a master or reference fingerprint etc. for c -rison. These c -rison means generate at least one ~ rison-result signal which is characteristic of the results of the c~mr~
son; and the previously mentioned signal which is characteristic of the relieved surface under test further includes this c _--ison-result signal.
When the system is to use known minutiae data for the particular specified relieved surface, the means for co~r~rison include some means for deriving new minutiae data, for the relieved surface under test, from the first-mentioned electrical-signal array. The means for c _--ison also include some means for evaluating the new minuti-ae data - relative to the minutiae data for the relieved surface.

W096/24283 PCTnUS96/01615 Both the minutiae-deriving and the minutiac e~aluating means are electronic processor means.
Preferably the deriving means comprise some means for processing the electrical-signal array by applying to ~t a direction map which 6 is characteristic of the Particular sp~ified relieved surface (that is to say, not n~c~s~:~rily of the unknown surface under test~. For breadth and generalit:y again, these means will be called the "prepro-reSSin~ means"-The preprocessing means preferably include some means for 10 applying a ~iltering algorithm to the electrical-signal array. For this algorithm it is preferred to use a mean-gradient filter.
In addition, pr~eferably the deriving means comprise means for applying a tuned filt:er, oriented in accordance with the direction map, to the electricaLl-signal array to obtain a filtered image. I
15 ~ss~nc~ such a filter can make use of already known information about fingerprints - such as the typical range of their ep~; ng - to ,_..~ve image noise and so ~nh~nc~ the ridge-and-groove patterns.
The deriving me~ans preferably further include means for ~nh~n~-ing binary character of the filtered image. Typically these provide some form of thresholding, but other t~hni ques are eq~uivalent.
It is prererabl~e that in addition the deriving means include means for applying a center and an axis of tne direction map to det~rmi n~ coordinates~ of the relieved-surface details. Advanta-geously the deriving means also include some means for using the axis 2~ of the map, and of the relieved sur~ace, to control the means for c~r~-ison - in part:icular to restrain the influence, upon finger-print properties that: should be topologically invariant and so upon identity-verification results, of grip-in~ll~e~ distortion of the finger as it engages the image-acq,uisition surface.
~ Incorporation into an ~c~ess-control system - Here the above-dis~-lce~ first: aspect of the invention is used to control ~c~cs to facilities, e~ir~~nt, a ~inanci~1 service or information, based upon surface-relief data from a relieved surface such as a 35 finger. In this case the functions of the electrooptical means are ~p~n~e~: they also apply the tet~i n~ identity to control ~cc~cs to the facilities, eq~uipment, fi n~n~i al service or information.

~ Incorporation into a secured system - A further preference 40 iS that the invention, ~n~r~ cs a greater system which includes, and is subiect to access control by, th~ ~ccecc-control system described in the pr~c~i n~ paragraph.

CA 02211910 199i-07-30 W096/2~283 PCTrUS96/01615 -28-Here the preferred form of the invention includes utilization means, susceptible to misuse in the ~s~nc~ of a particular relieved surface that is related to an authorized user. These utilization means are a facility, an apparatus, some means for providing a 5 f;n~nriAl service, or some means for providing information.

~ Incorporation into a personal weapon - In yet another pre-ferred form the invention is a personal weapon, subject to ~r~eS
control based upon surface-relief data from a human user's finger.
The weapon may be a firearm that includes means for dis-charging a bullet, which re~uires an enabling impact or other signal for detonation. Upon detonation a bullet flies to a target, trans-mitting energy to the target and thereby typically kills, disables, stuns, wounds, frightens, or otherwise infl~l~n~s an adversary.
More generally the weapon does not n~cess~rily fire bullets, but includes some means for discharging some energy-transmitting agency to thus influence an adversary. The agency-discharging means require enablement for their operation.
As will be understood this agency may take any of a very great 20 variety of forms, including but not limited to a laser beam, electri-cally charged barb, poison-carrying dart (in addition to the momen-tum-associated energy n~ for injection, poison for present pur-poses may be taken as representing a form of chemical energy), gre-nade, gas canister, soft block for crowd-control purposes, and so on.
In addition the weapon of this form of the invention includes in its entirety the data-acquiring-and-using apparatus according to the ~cc~ss-control form of the invention, as described above. Here the electrooptical means apply the det~rmi n~ identity to control enable-ment of the agency-discharging means.
This preferred personal-weapon form of the invention provides very significant advances relative to the prior art, but for greatest enjoyment of the benefits of the invention it is preferably practiced in conjunction with certain other features or characteristics which ~nh~n~e its benefits.
~ Finger bracing - For example, it is preferred that the weapon further include a h~grip for firm grasping by the human subject's hand to support the weapon. In this case a particular finger is braced ~g~in~t the first end of the prism means by the firm ~o grasping; further preferences in this regard have been described above in connection with the first aspect of the invention.

CA 022ll9l0 l997-07-30 W096/2~283 ~ /US961ol6ls ~ Unitary module - Another preference is that at least a part of the dischar~ing means and at least a part of the electronic means be physically formed together as a substantially unitary elec-tronic module. Alfl~o~h of course no such module can absolutely 5 prevent corruption or other ci~ tion of a security system by a detormin~ and e~ ful technician, formation of key components in unitary form does deter unauthorized byr~si ng of the fingerprint-data ~cc~s control.
For this ~L~ose preferably the substantially unitary electronic 10 module includes a unitary integrated-circuit chip carrying both a part of the ~i;Sr~h~rsring m~ans and a part of the electronic means.
Manufacture of intesrrated circuits at present requires relatively very ~Xp~ncive equipment that is accordingly available in only a relatively small number of facilities.
This intrinsic hedge a~i~ct high-te~hnology manufacturers may be ~nh~ce~ by deliberate design of the module to require the most advanced and costly e~i_ ~ t. For ~Y~r1e the chip and its avail-able space may be designed to require the highest degree of available microminiaturization, advantageously in conjunction ~ith extraordi-20 nary operating speed or other te~hnology-p~ching ~ -n~c Preferably too the module includes a shape that matches and is required by an electronic ...odule receptacle of the weapon. This shape includes electr~des ~or contacting elements of the weapon to effect the discharging. Preferably this shape includes complicated 26 contours to discourage the great majority of would-be ~po~ers.

~ Contro:Lled-impulse Ai s~h~ge - Moreover it is prefera-ble that the "at lea~st a part" of the discharging means include some means for providing .a epo~i fically controlled electrical imr--1c~ to 30 effect the discharging. Accordingly the energy-~isch~ging agency is manufa~Lu~ed to rocpon~ exclusively to the cp~rifically controlled electrical i _ ~1s~.
Spo~i~1 attention should be devoted to defeating efforts to by-pass the ~r~s-control system if the energy-~is~h~-ging means are 3~ relatively conv~rtional means such as bullets, as bullets carry with-in ~h~C~lves the mecms for their own propulsion. A hammer detona-tor, in the weapon, c:ontrolled by an opto~1~rtronic module is vulner-able to bypass of th6! ele~L~o.-ics, unless cp~i~1 measures are taken.
It is preferabl~e to provide a bullet that holds a charge of 40 explosive ~ow~r and an electrical detonator to ignite the explosive charge. The detonatclr in each bullet is manufactured to r~spon~
exclusively to the sp~o~;fically controlled electrical impulse -_ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ W096/24283 PCTnUS96/01615 which is to say, the detonator is manufactured so as isolate the charge ~g~inet other impulses, including for example a relatively high voltage applied willy-nilly in an attempt to fire the weapon.
Preferably the detonator responds exclusively to an impulse 5 whose duration or voltage is in a narrow range, or having a particu-lar waveform, or conveying specific information - or preferably to an impulse characterized by all four of these control parameters.

(b) A s~c~n ASPECT of the invention - In a s~o~ of its 0 facets or aspects, the invention is apparatus for acquiring surface-relief data from a relieved surface such as a finger.
The apparatus includes a prism which is formed from a bundle of optical fibers. This prism in turn includes:

~ an optical data-input end for contact with the relieved surface at the fiber t~rmin~tions, ~ at least one side face for receiving light into the prism, and ~ a partially reflecting surface for redirecting light received through the side face to illuminate the data-inpu' end.

The optical data-input end includes t~rmin~tions of fibers of the bundle. The redirected light illuminates these fiber tD~min~tions at 26 the data-input end.
In addition the apparatus includes some means for receiving light that p~es~s through the partially reflecting surface from the illuminated fiber t~min~tions, and in response providing at least one sign~l which is characteristic of the surface relief.
The foregoing may represent a description or definition of the s~co~ aspect of the invention in its broadest or most general terms.
Even as so couched, however, this second aspect of the invention resolves the problems of the prior art, by providing a simple and 35 elegant light-injection system that illuminates a finger applied to a fiber-optic prism - without any of the marginal operation or related deficiencies of the prior art.
Nevertheless to optimize its perform~nc~ this s~o~ aspect of the inven.ion is preferably practiced in conjunction with certain 40 advantageous characteristics or features.

W096/24283 PCTrUS96/01615 (i) Light-projecting means - For example it is preferred that the apparatus further include some me2ns for projecting light into the light-receiving face of the prism.
These "light-p;rojecting means" function so that the light crosses part of the bundle to enter individual fibers, through their respective side walls im~iately adjacent to the partially reflect-ing surface. The light so entering the fibers is partially redi-rected by that surfa.ce toward the data-input end of the prism, for illuminating the t~rmin~tions at the data-input end of the prism.

(ii) Ret:urn of relieved-surface scattered light through the partial reflectc~r - Another preference is that light r~hing each data-input-end t~-rmin~tion, ~p~n~i ~g on contact between that t~min~tion and such relieved surface, can be at least fractionally p~es~ from that t~rmin~tion back into and along the fiber.
This light, fr,actionally p~e5~ back into and along the fiber, in turn passes fract.ionally through the partially reflecting inter-face and the second optical element to the light-receiving means.

(iii) Second optical element - In this preferred case the apparatus also includes an optical element formed from another bundle of optical fibers, secured to the prism along the partially reflecting surface a~nd including an optical data-output end for p~ec~go of light 'raLveling along the fibers from the data-input end and through the part:ial reflector to the light-receiving means.
Such a second element, interposed between the partial reflector and the light-recei~ing means, simplifies isoiation of the light receiver from the lish_ being injected by reflection at the partial reflector. The part:ial reflector is used in two stages of the apparatus operation - injection, and backscatter return to the receiver - and thosle skilled in the art will appreciate the poten-tial for crosstalk, particularly l~ ~ of the injection lignt to the receiver thus b~assing the data-input er.Ld of the prism.
The second ele~ent, however, at least in ~heory is not strictly 3~ ne~ee~ry as crosstalk may perhaps be avoided through other tech-iques such as pola ization cont-o; or injected light and just behind the partial reflector. If so, then the receiver could be applied directly to the bacX: of the latter polarization-control point.

(c) A THIRD ASPECT of the invention - In a third major facet or aspect, the invention is a personal weapon subject to ~cc~oss control based upon s;urface-relief data from a human user's finger.

CA 02211910 l99i-07-30 W096/24283 PCT~S96/01615 _32-~he weapon includes means for discharging an energy-transmitting agency to influence an adversary.
The weapon also includes some means for developing an electronic signal representing a det~rmin~tion of whether a human user's finger-5 print data which are presented to the weapon are a particular autho-rized user's fingerprint data - and for also applying the electronic signal to control enablement of the agency-discharging means.
The weapon further includes means for deterring unauthorized bypassing of the fingerprint-data access control.
~o In one preferred form, the bypass-deterring means include physical formation of at least a part of tne discharging means and at least a part of the developing-and-applying means together as a substantially unitary electronic module.
In a second preferred form of this aspect of the invention, the 15 bypass-deterring means include means, associated with the discharging means, for providing a specifically controlled electrical impulse to actuate the energy-transmitting agency. By "specifically controlled electrical impulsen is meant ror example an -mpulse having one or more of these characteristics:
~ duration within a relatively narrow range, ~ voltage within a relatively narrow range, ~ a p2rticular waveform, and ~ a data stream conveying particular information.
In this second preferred form, the agency-transmitting agency is correspondingly manufactured to respond exclusively to the specifically controlled electrical impulse. For example, the energy-discharging agency may be a bullet having therein a charge of explo-sive powder and an electric21 detonator to ignite the explosivecharge; in this instance the detonator is manufactured to respond exclusively to the specifically controlled electrical impulse.
In yet a third preferred form of this third major facet or aspect of the invention, the weapon also includes some means for 3~ providing entry to working parts of the weapon; and the bypass-deterring means include:

second means for applying the electronic signal to control enablemeni of tnese entry-providing means, means for det~rmi ni ng when entry to working parts of the weapon is ~ine~ without operation of the second means, and W096/24283 PCTrUS96/01615 means for substantially p~r~=n~-ntly disabling the weapon when entry to lworking parts of the weapon is g~i nr~ without operation of the second means.

5 These last-mentioned disaDling means are responsive to the entry-det~rminin~ means.
The developing-and-applying means preferably include a crosslit optic-fiber prism with numerical aperture not r,x~r,~ing one-half, for collectins fingerprint data by frustrated total internal reflection.

All of _he ~ore!going operational principles and advantages of the present invention will be more fully appreciated upon consider-ation of the following detailed description, with reference to the apr~n~ drawings, oE which:

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. l is a generalized block diagram of a preferred ~o~i~t of the invention, al~so showing in a general way the operation Oc a representative fiber prism;
Fig. 2 is an is;ometric or perspective view of the Fig. l prism;
Fig. 3 is a diagram, showing verv fine detail but not to scale, 2s representing the process of fiDer-prism crosslighting in cross-section - the plane of the drawing being normal to the fiber axes and thus the prism a~is;
Fig. 4 is a li~:e diagram in axial section, aligned on the page with Fig. 3 for coordinated discussion;
Figs. 5, 8 and ll, which relate to three different crosslit fiber-prism configurations, are drawn witn a f -n orientation and size oC the receiving surface - and with a c~mm~ orientation and size of a representative thumb or other relieved surface - to show 35 on a fair ~r~rative basis the ~omm~n~l ity between the three forms and in particular to show that the main difference between them is in orientation Oc the fibers in the prism means; Fig. l~ supplements Fig. li in snowing how light is in~ected into the Fig. ll system;
Figs. 6, ~, 9, lG, 13 and 14 relate to six different forms of the abovc -~tioned three crosslit fiber-prism configurations, and are drawn with tne finger-receiving surface at a - ~- orientation and size - particularly to show on a c~p~rative basis how tne CA 022ll9l0 l997-07-30 W096/24283 PCTrUS96/01615 -3~-illumination path lengths turn out (sr--~hat coincidentally) to be the same in all six configurations; and also to show the significant differences in overall system size;

Fig. 5 is a diagram, like Fig. 4, of bright-field FTIR image generation in a preferred embo~im~nt with a crosslit fiber prism;
Fig. 6 is a diagram, not to scale, of a bright-field FTIR fiber-prism optical module that is 2 preferred embo~ t for implementing the Fig. 5 process;
?O Fig. 7 is a like diagram of a preferred embodiment that is a similar optical module but with a taper;

Fig. 8 is a diagram like Figs. 4 and 5, but showing dark-field FTIR operation in a preferred embodiment with a crosslit fiber prism;
Fig. 9 is 2 diagram like Fig. 6, but showing a dark-field module with a rectangular prism, another preferred embo~;m~nt for im~lement-ing the Fig. 8 process;
Fig. lO is a like diagram of a preferred emboAim~nt that is a similar optical module but with a taper;
Fig. 11 is a diagram like Figs. 3 through 5, and Fig. 8, but showing dark-field image generation in a preferred ~mho~im~nt with axial lighting at the subject;
Fig. 12 is a like diagram showing a crosslit int~ m~iate 25 reflector that is a preferred embo~im~nt for use in conjunction with the Fig. 71 system;
Fis. 13 is a diagram like Figs. 6 and 9, bu~ showing a dark-field module with int~m~-~iate reflector, another preferred embodi-ment for implementing the processes C4 Figs. lO and 11;
Fig. 14 is a like diagram of ~ preferred em~bodiment that is a similar optical module but with a taper;

Fig. 15 is 2 left elevation o~ a personal weapon according to a preferred Qmbo~im~nt, w th portions of the case drawn broken away to -~s show schematica~ly the internal mechanical layout, and exemplarily having a taper and a detector array -~s in Fig. 7;
Fig. 16 is a rear elevation of a variant of the Fig. 15 em~bodi-ment, mostly in section taken along the line 16-16 of Fig. 15;
Fig. 17 is a right elevation of 2 variant of the Figs. 15 and 16 40 embo~i - t, drawn partly broken away to show portions of the left side too, and in the interior ~chematically a fingerprint-acquisition optical element as in the em~bodiment of Figs. 11 and 1~;

-3~-Fig. 18 is a functional block diagram of the Fig. 15 weapon;
Fig. l9 is a partial m~hAnical layout, showing very schemati-cally a preferred e~o~im~nt that is an alternative for a portion of the Fig. 15 mechanical layout, this alternative having a unitary pro-cessor/actuator modu]Le;
Fig. 20 is a partial block diagram like po-iions of Fig. l but including a variant of constructior. that corresponds to the Fig. l9 alternative;
Fig. 21 is a very schematic elevation of a unitary data-proces-0 sor ana projectile-actuator module tha' is a preferred embo~im~nt particularly ~or the Fig. 15 weapon as modified by Figs. l9 and 20;
Fig. 22 is a very schematic longitll~i n~l section of a projectile device, with integra:! detonator, that is a preferred emboAi~~~t for use with the weapon of Figs. 15 through 21;
Fig. 23 is a functional block diagram, or flow chart, primarily showing a secona level of detail within the "new-minutiae deriving means" of Fig. l, an~ thus representing firmware which is a preferred ~mho~im~nt of the processing aspects of the invention, for any of the same systems as in the earlier figures, including actuation of the personal weapon or o~her utilization ~eans; and Fig. 24 is a schematic representation cf part of a direction map as used in the Fig. :23 - ~o~i~nt o~ the invention.

OF THE PR~KK~ EMBODIMENTS

30 1. OVERAL~ SYSTEM

~ a) Relieved-surface imaae acauisition and verification for securitv - As shown in Fig. l, p-eferred forms of the invention inciude a fiber-optic prism l0 for contact with a relieved surface ~ 35 such as a finger 1~ ! to provide an image of the relieved surface to el~ctroopt.cal means 20. In this ~orm of the invention the previ-ously mentioned "prism means" comprise a unitary prism l0 as shown.
The electrooptical means 20 actuate ~ss-control means 30, to either enable operat:ion of utilization means 40 if appropriate 4~ authorizaiion is ~m~o~ed in the received image, or maintain the utilization means disabled otherwise. For r~AeonC explained in sectio~ 4(a) below, the apparatus preferably ~c~ Ates collection CA 022ll9l0 l997-07-30 W096/2~283 PCT~S96/01615 of surface information from a thumb 11, although in principle other fingers can be used instead.
The prism lO includes a first end 1 for contacting the thumb 11, and secon~ end 2 for transferring the image to the electrooptical s means. The prism 10 also includes a side face 3 for receiving light, preferably infrared light, to illuminate the thumb 11; the width ~im~ncion of this side face 3 runs in and out of the Fig. 1 plane, but appears in Fig. 2.
For optimum operation, as suggested by Fig. 2, the width of the side face 3 - that is, the distance W3 in the drawing - is the same as the width wl of the first end 1. This condition departs from the geometry in Dowling.
Associated with the prism lO are a light source 4, and ~ fiber-optic spacer element or other diffuser 5 to somewhat e~ualize the illumination at the near and far sides of the prism 10. In the embo~im~nt of Fig. 1~ for reasons that will shortly ~pp~r, the spacer 5 should also be a polarization filter if the source 4 does not inherently provide polarized light.
Light rays 14 from the source 4 pass through the optional spacer 5 and cross varying fractions of the prism thi~Pn~es, as shown, to reach the c~con~ end 2 of the prism. In so doing, the light must cross optical fibers 51, preferably fused, which make up the prism lO
and define the prism axis 8 (Fig. 2). The light 14 pACS~c into the prism lO at a steep angle - as understood from Figs. 1 and 2 a right 25 angle - to the axis 8.
Accordingly the light 14 is not ducted along the fibers 15 in r~Arhi ng the second end 2 of the prism. In particular this light must cross through the side wall of each fiber to reach the t~rminA-tion of that fiber which contacts the second end 2.
Hat~hi ng used in the drawings to represent the fibers is only illustrative, since the fibers are essentially microscopic. They are preferablv spaced at some one hundred or two hundred to the millimeter.
In the embo~im~nt of Fis. 1, light r~Arhing the C~csn~ end 2 or 3~ the prism lO encounters a partial reflector 6 that is formed on the end 2. The reflector 6 is in ~C-c~nc~ a half-silvered mirror, but the exact fractions of light which are reflected and transmitted are subject to design choice and not nec~cs~rily half.
A portion of the inc~i ng light i4 r~Arhi ng the reflector 6 is redirected to form rays 15 ducted along the fibers 15 toward the thumb 11. The ~. -i n~r of the inr~i ng light 14 pACs~s through the partial refiector 6.

CA 022ll9l0 l997-07-30 W096/2~l283 PCT~S96/01615 In the exemplary Fig. 1 embo~im~nt, such light traversing the reflector 6 strikes ~a detector 7 - part of the electrooptical means 20 mentioned above -- held behind the reflector 6. For rt~Conc that will become clear mornentarily, stringent measures must be taken to 6 mi ni mi ze this effect.
The reflected and ducted rays 15 flood essentially all the fibers 51 of the pri:,m, and with relatively uniform intensity, to illum~nate t~in~tions of the fibers 51 zt the ~irst end 1 of the prism 10 and so illuminate the thumbprint or othe- relieved suxface 10 11. By virtue of FT:~R relationchips described earlier - and de-tailed in section 2(~) below - this illumination either is reflected (at thl~mhprint grooves) out or rhe prism as exit rays 16 or is transmitted through ~some of the fiber t~rmin~tions into the relieved surface 11 (at ~l " rint ridges).
The relieved surface 11, and the mass of living tissue (or other material~ within or Tnt~hinti that surrace, is slightly translucent and acts as a scattering medium - diffusing and redirectir.g the incident light 15 in all directions though not uniformly. A small ~raction, perhaps on the order of one thirtieth, of the light fraction trans-20 mitted into this medium is partially scattered as rays 17 back into and along the same fibers 51 which brougnt the illumination 15.
The latter, backscattered light 17 thus exists only in certain fibers that are in effect selected by the geometry of the thumh or other relieved surface 11. These rays 17, and the pattern of their 25 oc~u~er.ce in some fibers 51 but not others, accordingly constitute the optical data or inform,tion signal which is collecied from the thumb etc. 11.
Upor. r~-~t~hing ~for the second time) the partial reflector 6, some of the light 17 p~c5~c through the reflector to reach the 30 previously mentioned detector 7. Most of the r . inint3 light (not shown) is wasted in reflection back toward the entry face 3.
Optical data 1'7 which reach the detector 7 cause the detector to generate correspo~t~ing electrical data signa;s 21, which are pro-cessed in later s~ages of _he electrooptical means 20. The optical ~ 35 signals 17, backscattered from the thumb, as pointed out above are much weaker than tne ill~mination 15 _ncident on ~he tnumb - and weaker still than the previously mentioned light i4 first rt~t-hing the reflector 6 dire~ctly from the source 4.
Therefore the Eraction of illumination light 14 which p~cc,~c through the reflectc,r 6 and reaches the detector 7 can readily over-whelm the fraction c,f the optical data 17 which likewise passes tnrough the reflector o to the detector 7. It is for this reason W09612~283 PCTrUS96101615 that measures are required to mi ni mize such direct passage of initial illumination 14 to the detector 7.
These measures preferably inciude orienting the polarization filter 5 to substantially exclude from the system light 14 that cannot be reflected toward the thumb ll, and providing another polar-ization filter (not shown) between the reflector 6 and detector 7.
This ~o~c~n is peculiar to : ',o~im~nts in which a detector is close behind the reflector 6 - namely, the Fig. l embodiment and a variant form of tne Fig. 14 : ~~o~im~nt, discussed in section 2(d) below.
Advantageously a multilayer interference coating of dielectric materials can be substituted for the partial reflector 6 and second polarization filter (not shown) discussea above. Such an interfer-ence coating is particularly effective in reflecting light that is polarized in one orientation but rejecting light polarized at right 15 angles to that orientation.
For this purpose, known systems for controlling the light that passes through a 45c prism in a polarizing beam splitter are suitable for use with the fiber-optic prisms of the present invention.
Generally such a system takes the form of a quarter-wave r~Con~nt 20 reflective multilayer dielectric s~ack, which can be coated on the 45c face of a fiber-optic prism. This sort of system is designed for use with monochromatic (or, more pr~is ly, narrow-spectral-band) light, such as radiates from the diode illuminators preferred for the present invention.
ReC~llce the diffusion of the light passing normal to the fiber axes is in a plane normal to tne fibers, and there is no or little scattering into other oblique planes, polarization parallel to the fibers is preservea. Unfortunately in tnis orientation the "p" wave is transmitted.
In the other orientation the "s" wave in c~mm~rcially available films is 98% refiected ir it has not been greatly dispersed, and less so if the direction of the ray is not in the direction of the initial collimation. The leakage of the illuminating beam into the detector will then ~ two percent - roughly equal to the light (of the 35 opposite polarization) that is scattered back from the contacting ridges, collected by the fibers and transmitted th-ough the polariz-ing films.
Higher contrast will require the use of further polarizing films, or an offsei fiber-optic plate with EMA as shown in Fig. 14 to separate the components by direction as well as polarization.

W096/2~283 PCTrUS96/0161S

Wit~in the elec:trooptical means 2G, electrical signals 21 proceed to ~o~r~rison means 22, which preferably take the fonm of a p ~ Gy r ~ - ~ digital p:rocessor. The comparison means accordingly comprise means for p~arforming the various ~r~ison functions, and 5 these function-perfo:rming means are th~m~ ves modules of the pro-gram, with correspo~in~ physical portions of the processor that execute the program ~nodules.
Majo- modules c,f the electrooptical means as shown include:

10 ~ new-minutiae deriving means 23, which analyze the electrical signals 21 to clet~-~mi n~ and transmit minutiae 24 ~but in a top-ologically near-invariant form, as will be detailed in subsec-tion 5 of this Detailed Description) for the subject ~humb 11;

~ storage 25 and supply 26 of reference minutiae for the thumb of a person who is authorized to nave access to the utilization means 40 (and, if desired, specific people who are to be denied ~o~ s); and ~ evaluating means 27 for comparison of the newly derived minutiae 24 against the reference minutiae 26 to generate a "go / no-go"
output sigr.al 28 or, if it is preferred, a more-elaborate actuating OUtpllt signal 28, to an access-control device 30.

~5 The latter device ~0 generates a control signal 31 which is of suit-able physical character for enabling operation of ~he utilization means 40. For example, for some applications a mechanical signal 31 may be required.

(b) Applications of the svstem - The utilization- -nc block 40 represents any of a wide variety or a~plications for an access-control signal 31 such as ~he present invention generates. One focus of this document is upon US2 of the invention in, or as, a personal 35 weapon; however, the invention is equzlly applicable to other appara-tus, facilities, fi n~n~ial services and information services.
The invention :is particularly suited to field applications that are extremely ~m~n~i ng in terms of overall microminiaturization and low weight, very short decision time with very high certainty and 40 reliability, and lo~ power. Personal weaponry is zn application which is particularly sensitive to several of these criteria, but W096/2~283 PCTrUS96/01615 close h~hi n~ are other portable personal devices such as cellular phon~C and so-called "notebook" computers.
Use of the invention to control access to public phones, auto-matic teller m~rhi n~c, and vehicle-usage ~cc~cs - even though much 5 less critical in terms of weight and power - all benefit significantly from the ~m~hility of the present invention to miniaturization without compromise of decision time, certainty, and reliability. In some uses, such as teiephonic and in-person credit systems, the apparatus of the present invention does not n~C~cs~ y 10 actuate a device to automatically grant e. a. credit, but can instead provide a visible, audible etc. signal to a human operator who then (possibly after considering other information and ~king decisions) actuates any neces-c~y devices.
Any or all of these means for utilizing the signal 31 of the 16 present invention are represented by the utilization means 40.

2. OPTICS

(a) Crossliahtinq an FTIR fiber prism - A mechanism for cross-fiber transmission and cylindrical diffusion is not generally recognized. Figs. 3 and 4 illustrate the basis of the effect, showing the elements in a fused-fiber-optic material highly magni-fied.
The drawings show a few relatively high-index fibers 51 - or fiber cores - in a lower-index-matrix medium. Such a medium is the result of fusing a bundle of clad optical fibers, in which the fiber cores 51 have hiaher index than the cl~i ng.
The cl~ g has merged together to form the matrix 52, leaving 30 only the original cores 51 to distinguish the positions of the original fibers. In such fused plates of the material, the periodic-ity D of ~he structure is typically six to eight micrometers - just slightly less than the full diameter of the original fibers, as a small amount cf the original diameter nas gone into filling the 35 hexagona~-closest-packing gaps between fibers. This document uses the word ''fibers" in referring to the o-igir.al cores in this fused structure as it will be understood Lnat they are all that rc--i n.c, in discrete form, of -.he original fi~ers.
Figs. 3 and 4 show light rays 14a-d traversing the material.
40 With respect to the radial plane of Fig. 3 ~normal to the fiber/prism axis), the rays passing through the material are refracted by the fibers 5', focusing and defocusing as they pass from one fiber to CA 022ll9l0 l997-07-30 W096/24283 PCT~US96/01615 -4l-another. R~l~C~ the indices are different at each interface there arise a very small fraction o~ reflected rays 14a', 14b', 14d' etc.
which tend to be diffusive - but only as seen in the normal plane of Fig. 3. The very small fraction so reflected ~r~n~c upon the ratio of refractive indices across each boundary.
In the axial plane of Fig. 4 the rays are not so de~lected and diffused but instead, preserve their angle of rise or descent, through both backward-direct.ed reflections 14a', 14b', 14d' and forward re-flections 14dN. Thus the effect is of a diffusion-like spread pri-10 marily in one plane and angular preservation in the other. The beamundergoes a slow continuous ~xr~ncion in the plane of Fig. 3, and the light difrusion slowly tends to become isotropic in the plane.
The model used herein to approximate crosslighting penetration is a one-dimensiona~. model that ignores the cylindrical cross-section 15 of the fiber. A sin~le expression ~or the loss at each transition from fiber to fiber is the well known expression for reflection due to difference of refractive indices, at normal incidence:

- = (ncore ~ n~ ) 2/ (ncore + n~ ) 2, (Eq. 8) If the mean thi~n~ss or distance between interfaces is D /2, and the reflective loss is regarded as a fractional attenuation ~ per unit length, this attenuation ~ per unit length can be set equal to sim~ly the reflection loss or Eq. 8 divided by that thickness D /~, a = r/~ D or 2r/D. (Eq. 9) This expression can be used in turn as the basis for a model of the attenuation that is an exponential decay, exp(-_a), where ~ is the same attenuation per unit length - often cal}ed an "attenuation coefficient".
Eq. ~ does not give an exact expression of the flux, but rather describes the basis of the diffusion over a distance which is short in terms of the diffuse-tr~n~ission characteristics of the material s - in particular, perhaps over a small range of about two attenuation lengths - after which other influences becom~ more important.
- Eqs. 8 and 9 c:an be used to develop another simplified expres-sion that shows how to optimize perfor~-nce of a crosslit prism.
Such an expression advantageously relates the refractive indices to the numerical apert'ure NA of the fiber bundle, using Eq. 2 - rewrit-ten thus:
(NA) 2 = ( core) 2 -- (ncladding) W096/24283 PCT~S96/01615 -~2-(-- core _ clad~ing) (n core + nc- ad ~ ng) (n core ~ a d ac~ing) ~ 2n avg. (Eq. 2') where _ avg i s the average of the two index values. Solving this for 5 the index difference a core ~ n cla ~ ing yields n core ~ n cladding = (N A) 2 J2 _ av5 (Eq. lO) Eq. lO, giving the lndex difference in terms of numerical aperture and average index, can be related to the expression given earlier for the reflection, Eq. 8, if Eq. 8 is similarly rewritten in terms of the average index _ avg (--ccr~ -- n~l~ln~n~) 2/ (2navg) , (Eq. 11) and then Eq. lO is inserted into Eq. 11, r = ~NA] 2 ~2_ avg ) 2 t (2n avg ) 2 (N A/2 n avg.) (Eq. 12) This allows rewriting the attenuation-coefficient expression of Eq. 9 in turn as -0: = 2 (NA/2navg)4/D, (Eq. 13) so that the light flux at each depth x goes as exp(-_~) or exp (-2 _ (N A/2 n avg ) 4/D) which will be recognized as the function in Eq. 1. Moreover, setting this expression equal to 1/e - which is equivalent to setting the a~_ lated attenuation equal to unity, _~ = 1, in Eq. 13 - yields an expression for required numerical aperture N A to make any given penetratlon distance x be exactly one attenuation length, 3~ _~ = 1 2x(NA/2navg)4/D
NA = 2 n avg(D /2 _ )lt4 (Eq. 14) To obtain ~his degree o~ penetration or better, at a distance XF
40 along the illumination path ne~ to reach the far side of the prism, the condition on numerical aperture h~c~ c instead =

W096/24283 PCTnUS96101615 -~3 NA ~ 2n~(D /2_~)l/4 which is Eq. 6. Just: one material c~ .ly on the market has an adequately low NA va].ue, namely 0.35, am~ly lo~ for the condition described algebraically above. This fact is shown by Table 1, which = _-~es l/e attenuat:ion lengths for some available materials - the o attenuation length L being found from Eq. 14 for a given combination of glasses as:
_ l/~ D(2navg~/NA)4/2~ ( q. 15) In Table l, _ represents the reflection at each interface as before.
Cl~in~ index ~s no1: always readily avzilable from manufacturers, but in Table l is ta1:en to be the same for all three glasses. The attenuation length L = 23.7 mm shown for NA = 0.35 agrees r~or~hly 15 well with a value of 20.2 mm observed in development of the present invention. Since the rough calculation ignores off-center rays subject to angled scattering, for which reflection will be greater, the lower observed ai_tenuation length (higher attenuation coefficient ~ makes sense.
To obtain some indication of tne limits of practical operation for various values of numerical aperture, some calculations have been carried to two attemlation lengths - intensity dropping to l/eZ.
A l/e2 intensit.y drop, over the length o' the illumination path to the far side of a unidirectionally lit prism or to the midplane of a prism with opposed lighting, is regarded as a limit of practical operation. It corresponds to intensity variation of nearly Table l: attenuation lengths for available materials, calcu-lated using 8-micron center-to-center sp~j ng coré cl~i ng average ~ = l/~
NA index index index - (mm) l 1.819 1.67 ~.008 0.5 0.66 l.657 l.5~ i.589 O.OOl9 ~.15 0.3~ l.56 ~.5~ 0.0002 23.7 ninety percent (of the incident intensity) across the prism for a unidirectional-lighting system and seventy-five percent for opposed.
These calculations have explored the numerical-aperture require-40 ments for both l/_ and l/_ 2, and for prisms o' various widths - and also the numerical-apert~re performance for the two c-o -~ially W096/2~283 PCTrUS96/01615 available materials whose numerical apertures are either relevant in terms of prior art or intro~llc~ as pari of the present invention.
If a thumb is to be placed on a prism 10 as shown in Fig. 1, with the thum~b long axis in the plane Oc the drawing, then the length s (also in the drawing plane) of tne first end 1 of the prism 10 must be some fifteen to twenty millimeters - or a nomi n~l value of 17~2.
It is possible, however, to operate a system with the thumb axis (not shown) r~lnning in and out of the plane of the drawing - and in this case the length of the first end 1 can approach roughly 10~z mm.
10 Table 2 shows the results of these calculations.
The number of attenuation lengths, more than nine, shown in the top line of the table for NA = 0.66 is proDably far beyond the quantitative applicability of the algebraic expressions used in preparing the table, but does serve to suggest qual~tatively that the ~5 numerical aperture of 0.56 used in the Dowling devices is much too high for satisfactory operation.
The bottom line of the table conversely indicates that a numerical aperture of 0.35 can illuminate a 17~z mm prism within Table 2: Numerical aperture equired for 1/_ and 1/e2 perfor-mance in prisms of various widths, and perfo-~-nc~ for ~ -r-cially available numerical apertures, calculated using 7-micron 2~ center-to-center spacing intensity variation:
rawnllmh~r of mini~-m intensity as a width atten'n fraction of ,m~j of NA lengths prism along (mm) path unidir'l opposed 17~2 0.66 9~ 0.0001 0.0002 0.42 0~ 46 2 0.i4 G.27 10.6 ,0.~0 0.36 17~2 0~ 38 1 0.37 0.65 10.6 0.42 17~2 0.35 0.8 G.43 0.73 4s the reach of less than one attenuation length. On the other hand, numerical aperture of 0.35 is too low for good ducting, and this fact W096/24283 PCTrUS96/01615 -4j-in turn leads to a need for caution in configurations that require ducting: as suggested in the Summary section earlier, use of sepa-rate elements for tapers or curved light pipes resolves this concern.
The three : ,hlci2ed values of numerical aperture in Table 2 have been used in de~ining various preferred ~m~o~i~-~ts o~ the invention, as expressed in the earlier S ry section and in certain of the ~pp~n~ claims. The foregoing analysis together with Table 2 illustrates the nature of the criticaliry o~ these values.
As Table 2 shoi~s, optical-fiber prism materials with numerical 10 aperture up to about 0.5, were they a~ailable r~mm~rcially now, would correspond to an operationally marginal selection of material; and materials with numerical aperture of 0.3~ tG 0.42 - or more gener-ally Sp~i ng about 0.4 - might be seen as providing an ideal tradeoff between ducting ability and ~m~n~hility to crossiighting.
Tighter ranges may b~e stated, based cn ihe tabulation, for prisms of specific width.

The illumination-path lengths used implicitly in calculating the variations given above are not equal to the prism widths. They will 20 be discussed in the subsections that follow.
Overall no~llnif'ormity of illumination across the finger-contact-ins face of the prism is suggested in m2ble - by stating the minimllm intensiiy as a fraction cf the mA~imll~ intensity.
Another possible notation is to state the excursion a~ove and 25 below a median value, as a percentage of that m~i~n - yielding for example values of +0.4999/0.4999 or +100% at the top of the right-hand colum~ and a ve:ry modest value o~ +0.135/0.865 or +16~ at the bottom. Table 3 states the overall variation according to this notation, and also d:isplays the fractions of incident intensity from 30 each side that should be observed at various distances across the prism first end 1 - and correspo~ g dist~n~s for a prism with opposed lightir.g, to be discussed shortly.
In the opposed-lighting case, incident intensity from each side is assumed to be hal~ tha' frGm a single lam~ in the unidirectional 3~ case. Also for the opposed-lighting case only the gua-ter- and half-way points are tabulated, since the pattern is symmetrical - the - three-quarter and "full" points are the same as the zero and quarter points respectively.
As shown by the column just to right of the "NA" figures, light reaching an entry face from the lamp on the oP~osite side of the prism contributes to intensity next to the entry face quite signifi-cantly (9/50 = 18% for NA = 0.35). Thus the tailing-off intensities CA 022ll9l0 l997-07-30 W096/2~283 PCT~S96/0161S
-~6-beyond the midplane are not at all negligible, and he}p very substantially to even out the overall lighting.
Table 3 directly shows how smoothly (as well as how minim~lly) intensity can be made to vary across the prism face, with suitable 6 choice of numerical aperture. It also directly shows how abruptly (as well as how greatly) intensity varies with a poor choice: even in the more-favorable opposed-illumination case for NA = 0.66, intensity falls to less than 0.01/0.5, or in other words less than 2%
of the initial intensity from each side, before the quarter-width 10 point is re~che~.
Other aspects of fiber-prism crosslighting that merit attention are the illuminating devices and their use. High-brightness light-emitting diodes (LEDs) appear to be the best choice in the present market: operating at 2 V, 80 mA a four-element array is inexpensive 15 and provides more than an adequate exposure in one sixtieth of a second. Such an array can be flashed from a 750 ~F ~p~citor provid-ing a 16 mW-sec pulse.
Near-infrared units are available, m~king operation of the device invisible. GaAlAs LEDs may be best, operating at 820 nm Table 3: Illumination variation for a 17~ mm prism, calculated using 7-micron center-to-center sp~ci ng 45-degree prism, rectangular prism, light from just one side light fr. BOTH sides frac.full-prism width frac.full wid.
va ~n NA var'n O,/4 1/2 3/4 full+ ~O 1/4 1/2 +
1O. 10 0.01 O.OC 0.00100 0.66 0.50 0.00 0.00 100 10.61 0.37 0.22 0.14 76 0.44 0.51 0.21 0.14 58 10.78 0.61 0.47 0.37 46 0.37 0.57 0.41 0.37 21 10.81 0.66 0.54 0.43 3~ 0.35 0.59 0.47 0.43 16 (where a CCD provides quantum efficiency of forty percent~, and providing continuous-wave radiant powe- at 1 mW. Other choices are 40 GaAs LEDs, which radiate efficiently at 900 and 940 nm, though at 940 a CCD's quantum efficiency is only seven and a half percent.
Some of the prism configurations intro~lc~ in this ~o~ .~t are subject to illumination hot-spotting near the side face of the prism CA 022ll9l0 l997-07-30 W096/24283 PCTnUS96/01615 -~7 means from which illumination enters. This problem is r~ i by use of opposed lighting, as shown in Tables 2 and 3.
Irregular illumination can also be addressed adequately by conc~ntrating the lis~ht - even by co~c~trating the lamps - di-s rected toward the far~ side of the prism means; or by providing a Sp~r~ 5 as already mentionedi or both. A spacer can be of fiber-optic prism material similar to that used in the prism means, or can be some other type o~ diffusing spacer.
Noncontinuous geometry of LEDs in a representative array may leave the system susceptible to a small amount of ripple in the lighting intensity. This does not appear to czuse any pro~lem.
More interesting are moiré patterns or other two- or three-r~i~ncionaL interference erfects apparently resulting from beats between the internal structure of the fiber-optic prism means (with a 15 period of about 0.19 to 0.2 mm) and the finger-ridge structure ~period of roughly 0.3 to 0.4 mm). Subtle h~nr~ n~ can result, but this too is adequate:Ly mi ni ~i zed by use of a fiber-optic spacer 5 so that the prism does not effectively form a sharp end at the illumina-tion -ace 3.
Typically h~n~i n~ and ripple effects are slightly more signifi-cant in the bright-field than in the dark-field case, because the data-processing system can misinterpret as ridges the intensity structure in the modulated bright area. This effective additional noise term forms sti:Ll another basis - one that is peculiar to 25 fiber-optic-prism systems - for preferring dark-field operation.

(b) Bricrht-field systems - As Fig. 5 suggests, the diameters of the fibers 151 in a fiber-optic prism ~TIR system are preferably 30 smaller than the rids~e spacing. Actually the fibers should be even smaller than the dra~ing indicates, in relGtion to the ridges, but this dr~wing suffice-; to illus'rate the p-inciples of operation of the FTIR fiber-optic prism of the present invention.
As mentioned before, the fibers 151 were originally individual 35 cores of discrete optical fibers, and the matrix 152 was originally discrete indiviclual cl~ i nq~ formed as annula~ cylinders, of those - discrete fibers. In the fused material the refractive indices of the fibers ans matrix are about 1.56 to 1.58, and 1.52, respectively.
A thumb 11 or other relieved surface of interest is pressed t the first-end glass surface lO~ of ine prism means llO, so that thl~mhprint ridge areas contact some of the fiber 151 termi-nations - for instance at one normal 154 to the glass surface lO1 -CA 022ll9l0 l997-07-30 W096/2~283 PCTrUS96/01615 while cavities 153 are formed between the thumb 11 and glass 101 by thllmhprint y~OOv~ areas elsewhere, as at another normal 154'.
The refractive index of the thumb skin and flesh is typically about 1.41, defining a critical angle of 68~ ~g~inct the fibers 5 (index 1.56) of the prism. Within a fingerprint ~-oov~ cavity 153, air is present - and its index of 1.0 defines a critical angle of about 41~ ~g~inct the fibers.
Small quantities of liquids such as water, oils and sweat are usually found in fingerprint structures, and particularly in and near 10 the areas of contact between fingerprint ridge and glass surface.
Drops 156 of these materials have refractive indices generally different from those of both the skin and the glass - for instance 1.33 for water, defining a critical angle of 61C Ag~in~t the fibers - and they participate in optical interactions at the surface.
Tnus the critical angles for the thumb 11 ridges and liquids 156 associated with them ~x~A 45~, whereas the critical angle for the thumb 11 grooves 153 ~s less than 45C
Representative light rays 114a and 114b both strike the glass surface lO1 from within the prism means 110. One of these rays 114a 20 r~h~e a fingerprint ridge region at one normal 154, and the other ray 114b r~ch~ a groove region at another normal 154'.
Both rays sirike the surface lO1 at approximately 45~ to the respective normals 154, 154' - less than the critical angle for ridges and their associated liquids, but greater than that for 25 grooves. A ray 114a which strikes the ridge region accordingly is not subjes~ to total internal reflection, although in general it can partiallv reflect (not shown) from the surface - and it partially passes tnrough ihe surface and into the material of the finger (and associated liquids).
Some of this transmitted light is promptly scattered as rays 114a' p~csi ng further into the thumb 11. Next, some Oc these rays 114a' in turn after penetrating some distance are rescattered or redirected along ray paths 155. For simplicity of the drawing, only one generation of such a rescattered or redirected ray is illus-35 trated, but in general actually multiple generations Oc such events occur - which is to say, some of the light bounces about within the thumb 11 extensively.
A relatively small nll-~r of the redirected (or reradiated) rays 155 in turn may reenter the prism as shown. The single reentrant ray 155 shown in Fig. 5 is ~rawn a' too steep an angle to be ducted along the fiber 151, and so represents light 155, 116 ejected laterally CA 022ll9l0 l997-07-30 W096/24283 PCTrUS96/~1615 across the fibers and thus wasted from the system: this is what h~rp~ne to most of those few redirected rays that reenter the prism.
It is important to notice that the direction of the fibers 151 - oriented at 45~ to the surface normals 154, 154' and at riqht 5 anqles to the incidenl: ravs 114a, 114b - f~rst enters into the discussion only at this point. The reason the reentrant ray 155 cannot be ducted alons~ the fiber 151 in Fig. 5 is that in this drawing the orientation of the fibers 151 has been ~hos~ to exclude such rays. In other ~ords, there is nothing inherently unductable about the ray 155 itse~lf, apart from its relative orientation with respect to the fiber 151.
A statisticalIy small number of other reentrant backscattered rays (not shown), howe~ver, do fall within the acceptance cone of an adjacent fiber 151 oriented as shown, and are ducted to the detector.
In a bright-field syst:em such as under consideration here, such statistically occurring rays decrease the contrast between ridges and y~o~ves - and for pu~poses of the present discussion may be regarded as one source of optical "noise" or h~kground.
Similarly other :eays 114a" are scattered directly ~that is, not 20 as reentran. rays 155) from the thumb 11 b~ck into the fiber. Most of these too are at an angle that is ~YC ssively steep for ducting along the fiber - and these also are ejected laterally, representing light 114aN, 116 that is wasted from the system.
Some relatively i-ew of the directly scattered rays 114a" are at 25 a sufficiently shallow angle to be accepted by an adjacent fiber 151 oriented as shown, and so are ducted as reflected rays 114al', 117 to the detector as shown. In this bright-field system these statisti-cally occurring rays r~epresent a second source of optical "noise".
A ray 114b which st-ikes the yL~ve region is subject to total int~rn~l reflection an,~ accordingly may fcrm an entirely reflected ray 114b', 117 trave~ing back into the adjacent fiber 151. This ray, traveling in its respe,-tive fiber from its surface normal 154', represents the desired optical signal.
In other words it: may be helpful to regard these -ays as data, namely the information that a ~oov~ is present adjacent to a partic-ular light-carrying fiber. It is these rays 114b', 117 which the fibers in Fig. 5 have been oriented to coliec~. Grooves in this system thus appear bri~ht, and ridges dark.

Now ~ing off from the microscopic view of Fig. 5 to the macroscopic picture in Fig. 6, a suitable prism means for bright-field operation may ~e as shown - unitary goO/45c prism llO. The CA 022ll9l0 l997-07-30 W096/24283 PCT~S96101615 -so-thumb 11 is applied to the hypotenuse 101, and the fibers 151 run parallel to one of the short legs - also identifiable as the light-entry side face 103.
Fig. 6 is essentially keyed to Fig. 5, the input light 114 from 5 the source 104 passing through the spacer 105 and side face 103 and across varying widths of the prism, and through the fiber walls to directly illuminate the fiber t~r~in~tions at the thumb-contacting surface 101. Resulting optical signals 117 are ducted along the fibers to the detector 107, which is secured to the output face or 10 "s~con~ end~ 102 of the prism.
Waste light 116 is mostly dissipated isotropically. Some slight image fogging at the detector 107, however, does result as suggested in the drawing by the orientation of the waste-ray arrow 116.
The crosslighting distances vary from extremely short, for light 15 originating in the upper right end of the angled source 104, to a ~~~i~ length of x F for the lower left end of the source. These illumination path dist~n~c can be seen directly in the drawing as the lengths of the three arrows 114 crossing the fiber structure 151.
From the 90O/45C geometry of the prism it follows that the ~ m illumination path distance equals the length of the thumb-contacting hypot~uc~ 101 divided by the square root of two - or in algebraic form _ F = LT /~
The configuration of Fig. 6, as can be seen by c~mr~ring it with the adjacent drawings at substantially the same scale, is advanta-25 geously very compact. It does, however, have some minor drawbacks.
One of these is that the illumination 114a striking the prismnear its left end is very close to the detector 107, so that waste light 116 from that region ~i. e., particularly scattered light from ~ rint ridges) is iikely to form a flare or strong fogging at the 30 detector 107 in that area. Another minor drawback is that the length of the thum~, applied along the hypot~nl~se 101 of the prism, appears foreshortened by the factor ~ at the detector 107.
Both these drawh~c can be eliminated by fabricating the prism in the form of a parallelogram, as indicated in the phantom line, 35 with ext~n~ fibers 151t' carrying the signal 117" to ~ longer detec-tor 107" that is parallel to the input face 101. This configuration, however, has its own dra~h~kc in some sacrifice of c~mr~ctness~ and more importantly in the added cost of '.he larger detector 107".
Nevertheless the alternative configuration 151N-117"-107n may 40 have merit in event of various possible favorable developments in detectors: .his is detailed in section 3 below.

CA 022ll9l0 l997-07-30 ., I
Some motivation arises to save cost by sornehow using a smaller detector 107. One way to ~cc~rlish this is with a taper 158 (Fig.

7) - transferring the thumbprint image from the output end 102 of the prism to a rnuch smaller detector 107' on the smaller output face 5 102' of the taper.
In this case th~a "prism means" comprise the prism and taper considered together. As the drawings show, the prism section in the c~uulld structure of Fig. 7 is substantially identical to that in the unitarv-prism case of Fig. 6; thus the illumination-path lengths 10 mentioned earlier are! the same.
To avoid severe foggirlg of the image in such a device, it is important to use a ta.per 15~ that is properly designed to adequately preserve ducting of the signal rays, particularly through the most strongly curved region of the fibers 159. This condition calls for a 15 fiber structure of relatively high numerical aperture, or use of EMA
material in the prism, o- advisably both.
Therefore - since the prism requires low numerical aperture and ~s~nc~ of EMA material - as rnentioned earlier the taper must be fabricated separately from the prism. Another reason for use of a separate prism now i5; that the materials co"v~r,tionally used in m~k; n~ low-NA fiber optics are not : - ~hle to the drawing t~hni ques used to make tapers.
It may be possible to ~ -c~te for anamorphic mapping within the prism section, by suitable design of the taper - e. a. with a tilted end section as is shown in Fig. 6 and a tilted detector 107N.

(c) Dark-fielcl rectanqular-~rism svstems - In Fig. 8 the illumination and thumb geometry are exactly the same as in Fig. 5, discussed earlier. Therefore the input illumination 214a, 214b and 30 resultant generation of directly and indirectly scattered rays 214a', 214a", 255, at the ri.dge normal 254, and generation of reflected rays 214b', 214b" at the y~oove normal 254', are all just as assumed for the bright-field apparatus.
In this systerr., however, the fibers 252 are at a different angle ~ 35 to the surrace 201 -- or in other words '.he prism surface 201 is cut at a different angle to the fibers 252. Here the fibers run normal to the surf~ce 201 and at 45c to the incident rays 214a, 214b (rather than the converse, _. e. at 45c to the surface and normal to the rays, as in Fig. 5).
Therafore a ~ifferent set of ejections and accept~c~s applies in this case. Here l:he incident rays 214a, are intro~l~c~ so that they cannot sirnply reflect to the detector; therefore in this case CA 022ll9l0 l997-07-30 W096/24283 P~~ '01615 the ridges are left to ~rp~r dark at the detector. It is a rela-tively large er of the ridqe-backscattered rays 214a", 217 - and 255, 217 - which the fibers are oriented to accept into their ducting mode, to form the optical signal.
Here therefore the data take the form of information that a ridqe is present adjacent to the illumination-carrying fiber. By ~ ison most of the FTIR groove-reflected rays 214b', 216 are ejected laterally from the system as waste light - with the impor-tant exception of certain rays 214b" to be discl~cc~ momentarily.
Except for the possibility of dirty or very shallow grooves, or backscattering due to ~Ccive liquids 256 in the ~h~mhprint struc-ture, there is relatively very little noise in this system. Hence in general a dark-field configuration tends to provide somewhat better contrast than a bright-field system.
As Fig. 8 shows, however, there is an important undesired phen~ -o~ in this rectangular dark-field system. Rays 214b" derived from the sp~c~llarly reflected ,-- _on~nt 214b' can propagate diffusely toward the detector by multiple reflections at the fiber interfaces.
These rays 214b" constitute, or are analogous to, a form of stray light diffusing toward the detector.
This propagation does not proceed by ducting of the rays 214b"
within the fibers 251, but rather by the previously dis~cse~ fiber-crossing diffusion process that is used initially to in~ect the illumination to the thumb-contacting t~r~in~tions. The best way to 25 avoid this effect is to absorb the stray light with suitably placed EMA material of proper numerical aperture.

Turning to the macroscopic view of Fig. 9, as can be seen the prism 210 is rectangular and considerably more massive than those of 30 Fig. 6 (although the length need not be quite as great as shown). Of cou-se the relatively greater mass in itself is a drawback for the most miniaturization-sensitive applications of the invention -particularly personal we~ro~c, portable telephones and the like.
In addit-on the illustrated auxiliary coupling prisms 205 add considerable undesired size and weight to the assembly. The function of these prisms is to avoid three problems: high reflection losses, low Lambertian (cosine law) flux through each side face 203 of the main prism 210, and difficult source-to-prism alignment.
All three would arise from direct approach of illumination 214 40 at a very shallow angle, and narrow aspect of only a few millimeters, outside the side faces 203. Antireflection coatings would address W096/24283 PCTnUS96/01615 -s3-only the first problem (reflection loss) and at such grazing inci-dence would be mi ni -lly effective.
This adverse ef~.ect, however, can be mitigated without ~ing such large coupling prismq 205. Mitigation is att~in~hle through -~ substitution of a '~r of smaller prisrns, side by side. While it is possible to use a large ~A~ of tiny molded facets similar to those of a Fresnel le,ns, an ideal nllmh~r is probably small, as for example between two and five prisrns.
T~n~; n~ to offset the draw~ of greater mass apparent in the 10 main prism 210 of Fig. 9 - as ~ ~-~ed with Figs. 5 through 7 - are higher contrast and also the improved uniformity of illumination discussed in section 2(a) above. ~ ce lighting can be applied from both sides some evenin~-out of the illumination is possible, and the overall profile becomes relatively uniform as shown by the right side of Table 3.
The principal cc,ndition of interest is the projection of light to the midplane. As can be seen directly from the ~ -n scale of Figs. 6 and 9, and the way in which these drawings are juxtaposed, the illumination path length XM to the midplane in Fig. 9 is exactly the same as the c~r~l~able distance XF of Fig. 6.
Light is supplied from twice as many directions - but only half as much light from each direction. Conc~quently a derivation of numerical-aperture constraints for crosslighting of the Fig. 9 system pro~ee~q to the same algebraic results, except for substitution of 25 the midplane notation XM for the far-side notation XF.
If desired, it is theoretically possible to halve the amount of light recluired~in this system - by omitting the inho~d half of each source 204. Such changes also enable use of half-size couplers 205.
Also enabled in turn by this hypothetical modification would be 30 a half-length main pri.sm 210, since the second end 202 could be moved nearly to the point of impi n; --t of those rays 214 shown in Fig. 9 as r~i ng from the centers of the sources 204. Such a modification thus would very great].y relieve the size, weight and cost drawh~q of the Fig. 9 system.
These benefits would be obt~ined at the sacrifice of the benefi-cial uniformity of il].urnination noted above, but uniformity would - still be reasonably good, eclual to that for unidirect-onal illurnina- tion as given in the left half of Table 3.
This modification recluires careful alignment of the two converg-40 ing beam edges at the midplane. ~t is likely that the graded edges of the two beams can be feathered together satisfactorily.

CA 02211910 l997-07-30 W096/24283 PCTnUS96101615 Detector-size ~onc~ns mentioned above in ~on~ction with the Fig. 6 bright-field apparatus are even more troublesome here, due to the larger cross-section, by the factor ~, of the Fig. 9 rectangular dark-field prism (particularly at its second end 202). They may be 5 addressed by use of a taper 258 as suggested in Fig. 10, but the taper ~i~ncions too are larger, by at least the factor ~, than the analogous bright-field taper of Fig. 7.
The theoretical length- and weight-saving benefits of omitting half the source 204 widths, dis~l~c~ above in relation to Fig. 9, 10 apply equally to Fig. 10 - but only to the main rectangular prism (210 in Fig. 9) and couplers 205, not to the taper 258. Dimensions of the taper are set by the main-prism unchanged widths, both in the plane of the drawings and out of that plane, not by the main-prism r~Allce~ length.
Fig. 9 also shows the previously mentioned stray-light rays 214b" derived from the specular reflection 216 and diffusely propa-gating toward the detector - transversely to the fibers, through myriad reflections at the fiber interfaces. These are captured and 20 absorbed by a short section, just before the second end 202, of fibers 251' that have EMA material.
The short EMA section 251' can be either a separately manufac-tured prism block or unitary with the main fiber block 251, as preferred. It is desirable that the numerical aperture of the EMA
25 section 25;' be about the same as the main block 251: if the EMA
section 251' has higher numerical aperture, it will accept and trans-mit much or all of the stray light - thus defeating the purpose of the absorbing section 251'. If the numerical aperture of the EMA
section 251' is lower than that of the main block 251, then some signal rays 217 will be undesirably discarded.
As Fig. 10 suggests, the EMA section 251' need not be long -but again it must r.ot have higher numerical aperture than the main block. Thus in particular the taper 258, 259 alone would fail to serve the purpose of intercepting the specularly-derive~ rays, and 35 would instead - by virtue of its much higher angular acceptance - ~
transmit a larae fraction of that stray light to _he detector.

(d) Dark-field ~artial-reflector sVstems - In the microscopic view of Fig. 11 again the tl ' contacting surface 301, contours of the thumb ;1, and incident light rays 315a, 315b are exactly as in Figs. 5 and 8. The fibers 351, however, are in yet a third orienta-tion relative to the incident rays 315a, 315b.

CA 022ll9l0 l997-07-30 WO 96/24283 ~CT/US96/~1615 ~;
Here the fibers 351 are parallel to the incident rays 315a, 315b - rather than at 90c as in Fig. 5, or 45~ as in Fig. 8. In fact, here the fibers 351 c:arry the incident illuminating rays to the surface 301.
6 Although here crosslighting is employed, it does not occur next to the tl ' contacti.ng surface as in Figs. 5 and 8 but rather at a partial reflector 306 (Fig. 12) which is adjacent to an opposite end of the tl ' contacti.ng prism. This system will thus be recognized as closely related to the configuration first intro~l7~e~ in Fig. 1.

On account of the dif~erent fiber orientation relative to the incident rays 315a, 315b, a different selection of the directly and indirectly b2ckscattered rays 315a", 355 at a ridge-adjacent normal 354 will be able to e!nter the ducted mode of the corresp~; ng fiber 15 and travel toward the~ detector as signal or data 317. Generally cp~ki ng the h~e~ttered rays which could be collected in Fig. 8 are too steep for duc:ting and instead b~- - waste light 316 in Fig.
11, as can be seen by a detailed c~r~ison of the rays shown in ~ ~~ in these drawi.ngs.
In a generic sense, however, the ducted h~C~tter in Fig. 11 is equivalent ~nd proportional to that in FicJ. 8. It therefore on r~hi ng the detector. pro~l~c~c substantially proportional or ecruiva-lent electrical signals - but smaller, for reasons to be disc~qs~
momentarily.
At a ~hl~mhprint y-oove 353 in Fig. 11, internally reflected light 315b', 315b" is ejected from the system as waste light 316 at even ste~r~r angles 1han the corr~cpo~ g rays 214b', ~14b" in Fig.

8. Therefore the op1_ical signal-to-poise relation as evaluated at this thu~b-contacting end of the Fig. 11 dark-field system is closely c ,-~able to that of Fig. 8, and p~h~rs better.
Differences do arise, however, at the other end of the thumb-contacting prism - shown in Fig. 12. Here entering illumination 314a", 314b" is inciclent, in the crosslighting mode, on a partial reflector 306 - whicn by reflection forms ducted illumination rays 315a, 315b propagating toward the _humb end of Ihe prism. A signifi-cant fraction of the initial illuminaLion er.ergy is lost in the process, though cons:iderably less than half if the incident light is polarized in the plane of the reflections.
At this same in,terface, some of tne returned h~q~tter 315a", 40 317 passes through tlle partial reflector 306. This process too is subject to loss, roughly half.

CA 022ll9l0 l997-07-30 W096l24283 PCTnUS96/01615 The overall loss in two passes at the reflector 306 leads to lower optical-signal intensity - and thereby lower signal-to-noise ratio on account of increased significance of shot noise - at the detector. Merely as an example, i~ intensity is lowered by an 6 overall factor of 2~ in the two p~c5~c at the reflector, then signal-to-noise falls by the square root of this nllmh~r: a factor of roughly 1.6.
If a coupling prism, taper or polarization filter is present, the radiation passing through the reflector 306 forms con~i nlli ng rays 414a", 317 within the fibers 451 of that element. In any event the through-radiation eventually enters a detector 7 as in Fig. 1 - or, turning now to the macroscopic views - 307 as in Fig. 13, or 307' as in Fig. 14.

It will be noted that in Figs. 12 and 13 the thumb-contacting first end 301 is shown at opposite +45~ orientations, relative to the fibers 351, 451 - but also relative to the interface and reflector 306. This latter relatio~chir is more -ningful, corr~cpon~ing to different overall ch~p~5 of the thumb-contacting prism: either a triangle as in Fig. 13, or a parallelogram generally as in Fig. 17 (or Fig. 6, but of course with illumination geometry different from that in Fig. 6).
These two overall shapes are functionally equivalent, as far as image formation at the detector is ~onc~rn~A. One or the other, 26 however, may be distinctly preferred for --h~ni cal ~c~ tion of the prism means in any given practical device.
In one preferred '_'i - t, of Fig. 13, the detector 307 is diCpoc~ at right angles to the fiber axis and ducted optical data 317. This system is ~oderately ~ -ct, but is subject to image 30 anamorphism.
Correction may be obt~ine~ through use of one alternative 45~
detector pl~c~m~nt 307N,drawn in the phantom line in Fig. 13. This alternative is subject to almost as much fogging as the Fig. 1 con-figuration, by incident source light 314 passing through the partial 3s reflector 306, but also is ~m~n~hle to several different cures.
One such cure is the same one proposed relative to Fig. 1, i. e.
installation of a polarization filter (not shown) - which here may be provided at either side of the narrow fiber-optic sp~c~~ section 410". Other cures for incident-illumination 314 fogging of the 40 detector 307" include leng~h~ing of that sp~c~r section, or ~-king it with EMA material or high-numerical-aperture material as pre-CA 022ll9l0 l997-07-30 _5,_ scribed earlier for the ~aper sections - or any combinations of these several tactics.
In general, use of EMA material is preferred for the ~n~-stage element 410, 410", regardless of the shape of that element -5 although this preference is much more important for narrow-slice sp~c~rs 410" as suggested just above.
I~ the first-st~age prism section is formed as a parallelogram, a possihility mentioned above, then the anamorphism-correcting alterna-tive detector pl~ ~ t 307" and thin spacer section 410" of Fig. 13 10 may simply follow the opposite 45~ orientation of the ~ond end 302 and reflecto~ 306 - in much the same way that the illustrated detector and sp~c~r 307", 410n follow the illustrated C~ron~ end 302 and reflector 306.
If preferred, however, the ~p~ may be made with the same triangular shape as the main prism - so that the remote face 302' (the second end of the overall prism means 310) is laid over toward the right, in a 45c orientation (not shown) parallel to the first end 301. This orientation too will eliminate the ~na~rphism of the detector p~c~m~nt 307 shown in the solid line in Fig. 14. In this 20 case it is the entire dual-element prism means, rather than only the first-stage prism considered alone, that is shaped as a parallelo-gram; such a confisruration is relatively long.
Now as to the cr.osslighting and numerical-aperture constraints requ-red in ~ny of these variants of the system of Figs. 11 through Z5 13, the m~j illumination path length ~pe~rs in Figs. 13 and 14 as the longest illumination ray arrow 314. Once again the .
scale and orientation of the ~- contacting first end in Figs. 6, 9 and 13 enables visual confirmation that the m~Yi ~ illumination path length x F~ identified in Fig. 13 is exactly the same distance as in 30 the two ~mho~imQnts dis~cs~ earlier - namely _ F' = LT /~ There-fore the previously intro~l~c~ calculations and tabulations apply equally well to Fig. 13 and its variants.
As with the bris~ht- ar.d dark-field embo~i ts of Figs. 5 through 10, to the extent that detector cost is an obstacle some L 35 relief may be obt~i n~ through use of a taper 358 (Fig. 14). R~c~l~ee the Fig. 13 first-stage ~rism does not present a s~on~-end face that is perpendicular to t]ne fiber axis, however, the taper 358 in this case is p~rticularly :long, ~-~ing the overall ~es-~ly somewhat unwieldy as seen in Fig. 14.
The added length may be alleviated by a variant (not shown) in which the crosslit lo~w-numerical-apertur~ element is at the narrow end of the taper inste~ad of the broad end. An interface between the W096/24283 PCTrUS96/01615 two elements may then be cut off square, and the angled partial reflector placed at the other end (i. e., remote from the taper) of the crosslit element.
In such a variant the finger-contacting end of the ~qs~mhly may 5 be continuous with the broad end of the taper, and of the same high-numerical-aperture material since the crosslighting is done at the narrow end. The detector is coupled to the crosslit se_ -t, either through an int~ ~'i~te sp~c~r (whicn may be either triangular or thin-slice-shaped, analogously to the prism 410 and sp~c~r 410" in o Fig. 13) or directly h~hi n~ the reflector/filter as described earlier in conjunction with Fig. l.
In this variant, as in the rectangular-prism configurations of Figs. 8 through lO, it is essential to control lateral diffusion of the sign~l rays transferring from the high-numerical-aperture taper to the low-numerical-aperture crosslighting section. ~ec~l~s~ of the much narrower acceptance angle of the crosslighting section, a significant fraction of the signal will be rejected from the ducting mode in that section.
This rejected light, however, will continue into that section by transverse diffusion, and if not controlled will fog the image at the detector. This can be prevented by use of EMA material in the final section - the stage between the angled partial reflector and the detector. Although in this case some signal is sacrificed in the EMA
material, the EMA material is desired anyway - and ne~ if the 25 final section is a thin-slice spacer - to prevent diffusion toward the detector of the initial input illumination.
The variant under discussion is shorter than the Fig. 14 embodi-ment, h~ll5 the taper is shortened by the length of the illumina-tion entry face 303' - and this saving is only partially c~ ~-~rq~ted 30 by the length taken up by the equivalent diagonal with its associated entry face (not shown) in the smaller-cross-section crosslit element.
The diagonal across this smaller crosslit element is shorter by a factor equal to the ~ gnification M in the taper, so that the over-all ~-ss~mhly can be shortened by the distance (1 - M~ x F~

3. DETECTION
I

(a) Geometrv and --h~ni cal arranqements - In principle, the 40 image pro~l~c~ by the crosslit fiber prism can be detected by using a lens to directly image the ouL~u~ face of the prism onto a self-5~nn~ detector array. Such a detector array, an array of photosen-W096/24283 PCTrUS96/01615 _59 sitive cells, convert-; the light distribution into a corresps~;n~
distribution of photoelectrons, which are then read out in series by some means of scAnnin~ the photosites in the array.
Although the present invention can be thus practiced using a 5 conventional imaging ~system, its preferred use is in an all-solid noni ging fiber-optic: system in which the detector is affixed by means of tr~nCp~rent optical means - epoxy cements, or index-match-ing oils with mechanic:al at~ nt - to the output face of a solid fiber-optic element. In order to prevent pattern interference and 10 moiré patterns, the si.ze of the photosites should be substantially different (prefere~bly larger) than the optical fibe~s, at least by a factor of two and preferabiy more.

(b) Cost tradeoffs with taPer - The present price of even a 15 relatively small detec:tor 107 such as shown in the solid line of Figs. 6 and 13, if inç~lemented as a co--v~L,tional chargc co~led detector (CCD) array, is high enough to constitute the major cost element in apparatus according to the invention.
This is the motivation for considering tapers 158, 258, 358 even though a taper in turn disadvantageously adds to the weight, size and cost of the apparatus. At this writing, however, the CCD cost advantage in provision of a taper 158 in many cases is more than offset by the ino-_.~ntal cost of the taper - even without consider-ing the weight and size penalty.
It is not possible to predict reliably whether eventual cost relief should be expec:ted in the detector or the taper, or neither.

(c) Detector t~es - Various possible developments could relieve the tension discussed above. The price of conv~ntional crys-talline-silicon CCD arrays in this size range may fall - perhaps partially in r~spo~ce to com~etition for usage in apparatus according to the present invention. An alternative optical detector, such as for instance a self-s~n~ diode ~"SSD") arrzy or ~he thin-film (noncrystalline) photcc~nqor array mentioned earlier, may h~
available at significantly lower cost.
If both CCD and taper prices remain elevated, and alternative detector devices are not ~ -rcialized, the result is to limit the personal-ea~ t applicability of the invention to the high end of the market - e. a., for personal weapons, the market for personal 40 w~p~nC costing perhaps $500 to $1,000 in 1995 terms.

CA 022ll9l0 l997-07-30 W096/2~283 PCTnUS96/01615 One way to correct prism anamorphism, imaging asymmetry, is to use a detector with complementarily asymmetrical pixels. Some relatively in~r~cive CCD arrays h~rp~n to be available with rectan-gular pixels, e. a. the Texas Instruments Model TC211, in which each 5 pixel is 13~ by 16 microns. Unfortunately the ratio of these sides is 1.16, not very close to the ~ ratio of ~ = 1.41.
We know of no - -rci~lly available unit that provides the desired ~ ratio - but as will be clear sucn arrays can be readily manufactured as high-volume ~ n~ appears, and presum.~bly manufac-tured in the same smal; size as the TC211 and at favorable costs,~ _-rable to that of the present TC211. Such made-to-order detec-tors would correct ~n: -rphism and under present market conditions would be less expensive than full-size CCDs, even t~ing into account neC~ss~ry added cost of mat~hing tapers - and therefore are the most 16 highly preferred detectors for use as part of the present invention.

Detectors that can be used are primarily CCDs (charge-coupled devices), CIDs (charge-injection devices), and SSDs (self-s~nn~
diodes). All of these are made in two-~ cional arrays by many 20 manufacturers: Texas instruments, Fairchild, Tektronics, Kodak, Dalsa, Phillips, Thnmcon, Sony, Hitachi and so on - and in a large variety of sizes, and costs.
Smaller and cheaper devices include the Texas Instruments TC211, with 192 by 165 pixels, m~Cl~ring 13~ by 16 microns as mentioned 25 above (for overall ~i cions 2.64 mm square); and the TC255 with 243 by 336 pixels m~Cllring 10 by 10 microns (overall 2.4 by 3.4 mm).
These are made for mass-pro~ consumer items and cost under $25.
They are applicable, and bnc~llce they are so e~o~n~ical are preferred, for the configurations using a fiber-optic taper with high 30 ~ nification; these are seCon~ in preferability h~hin~ the custom-sized units already disc~lcs~. Most detector arrays now available are 8.8 by 6.6 mm; these could be used but are currently about $100 - and also still call for a taper.
Larger 512-by-512 (and greater) CCDs and large-pixel SSDs are ~s applicable and preferred for the direct-transfer configurations; they are third in preference, behind the high-volume units mentioned in the prec~in~ paragraph. If cost b~cnm~s cnm~rable, however, they could h~c more preferable - in view of the overall ~nm~ctness which they would confer on the system, by virtue of n~i ng no taper.
~o A Rodak Model 4000, which has a 4096-by-4096-element area, is big enough to use witnout a taper, though at present far too costly.
Al~hol~gh the CCDs are extremely ~xr~ncive in these large sizes, the CA 022ll9l0 l997-07-30 SSDs can be made Wit}l in~r~n~ive materials such as amorphous silicon and may enable lower overall system cost.

b 4. PERSONAL-WEAPON IMPLE~ENTATIONS

(a) Mechanica]. arranqements and function - A personal weapon according to preferre-d embodiments of the present invention may take the form o~ a pistol (Figs. 15 and 16j having a barrel 41 and a 10 firins ~mh~r 43 for holding a bullet 43 in firing position.
A generally conventional magazine 45 within the pistol handle 46 stores a n~h~r of aclditional bullets 44' for f~i n~ into that firing position in se!q~uence as bullets are used. A trigger 42 is provided for generally conventional operation by a user, but as will 15 be seen the action of' the trigger 42 upon the functioning of the pistol is preferably not conventional.
Exposed at upper left on the handle 46 are a thumb rest 61 and, ~h~ in the thum~ rest, the first end 101 of a prism means 110'.
The prism means 110' are held within the left side of the handle and comprise, in addition to the pri~,.,...~ans first end 101, a taper 158 generally as in Fig. 7. Shown schematically within the handle 46 are a CCD array 107', circuit boards 63 and electrical batteries 64 for powering the lamps and electronic system of the invention.

2~ Fig. 17, while showing a variant 501-507 of the prism means and detector, primarily illustrates a novel ergonomic pl~c - t of the prism, which makes fingerprints acquired with the apparatus signifi-cantly more distinct and much more repro~ hle.
The only currently known prior representation of a fingerprint-controlled personal weapon indicates pictorially that prints are tobe taken from somé finger or fingers other than the thumb, and by means of finger pressure where the finger wraps around the handle.
That prior disclosure! makes no provision for repro~rihle position-ing, orientation, or pressuring of the finger ~g~ t the weapon handle - or for ci,.~,,~e--ting the user's failures in this regard.
Many of the otherwise well-elaborated inventions in the auto-mated fingerprint-acquisition art, apart from personal-weapon con-trol, are similarly lax. Resulting inCOnCiStencies in input data lead to greater p~vy~.. complexity and more time, to obtain any ~p~ified level of identification certainty.
From study of Fig. 17 it will be clear that the iilustrated gripping arrangement greatly re~llc~s these repro~ hility problems.

CA 022ll9l0 l997-07-30 W096/24283 PCTnUS96/01615 Natural grasping forces A and B of a user's palm and fingers, respec-tively, ~g~inct the handle 46 coop~rate with instinctively applied force C of the user's thumb ~g~inct the prism first end 501, to brace the thumb 11 ~g~inct the prism.
This bracing action places the thum~ within a very narrowly constrained range of positions and orientations - and also stabi-lizes the thumb 11 in this position and orientation ~g~i~ct the prism surface 501. In addition this gripping geometry and bracing tend to make the degree of pressure of the thumb 11 on the prism relatively 10 consistent: an important refin~m-nt of the invention.

Fig. 18 indicates the functionality of the items in the ~h~ni_ cal layout of Figs. 16 and 17. Preferably the trigger 42 when depressed 68 turns on 66 the battery 64 conn~ctions 65 to the system, 15 enabling 67 illumination of the s~ci~g face 1 by the diode array 4.
Con~l~rently, powering up the system sends timed excitation 29' from the CCD driver 5, 29 to the CCD 7, and strobe pulses 29N which meet the CCD data 2' in the firmware-controlled processing stage -particularly to enable new-minutiae derivation 23. Resulting minu-20 tiae data 24 proceed to the FPGA/system block 27/30, for c~r~isonwith stored 25 authorizing (or deauthorizing) identification 26.
The FPGA/system block 27/30 in turn controls 31 the energy-discharging means 40 - here t~ki ng the form of the bullets 44.
Advantageously part of the battery system 64 is a powe~ n~gement 25 function, which receives fed-back information 65' about power con-sumption, electrical shorts etc. and applies this information to limit or ' l~te power to the rest of the apparatus.

(b) Unitarv Processor/actuator module - A system such as shown in the general block diagram Fig. 1 is particularly vulnerable to tampering or bypassing at two points. One is the evaluating --n.c output 28 to the ~cc~ss-control block 30, and the other is the ~c~sc-control-block o~L~uL 31 to the utilization means 40.
At these points an entirely straightforward system design would provide relatively simple "go / no-go" control signals. Knowledge-able but unauthorized individuals wishing to defeat the identifica-tion provisions in the ~r~ison means 22 may seek to interrupt the circuit at such points. For example such an individual may wish to 40 substitute always "go" control signals and so enable use of the weapon (or other secured system) without authorization.

W096/24283 PCTrUS96/01615 Fig. 20 shows schematically a modification of the Fig. l system that can deter, thousrh not absolutely preclude, such unauthorized use. A person havincr adequate resources, persistence and information about the control sy~;tem cannot be absolutely precluded from using 5 the weapon - regard]ess of what deterrence is built in - but the present invention se~es its purposes if such use is made very slow, troublesome and generally lln~cQn~ic.
The Fi~. 20 system attempts to eliminate the points of w lnera-bility, by l; nkin~ the successive adjacent blocks at the two sides of 10 the vulnerable point;. The evaluating means 27 are linked with the Acc~ss-control block 30, and the ~ccess-control block 30 in turn is linked with the utilization means 40. The originally vulnerable points are buried wit:hin the l; n~g~ functions, ; n~cc~csi hle or at least e~L~,~ly resistant to tampering.
The 1 i ~ki ~g may be ~ lished physically, as by building all or portions of the two blocks 27, 30 as a physically ine~p~able and unitary module; or f~mctionally as by so intertwining the operations of the two blocks 30, 40 that neither can function without the other;
or both. In the case of the utilization means 30, since it is 20 natural to make these physically separate objects - partially projectiles, in fact - it ~pp~s ne~ee~y to rely upon functional interdep~n~nc~ exclusively.
Figs. l9 through 21 show s~ hat schemaLtically how a hybrid physical/functional li nki ng of adjacent blocks may be provided. For 25 eco~y, the bulk of the new-minutiae deriving means 23 and other hardware of the invention may be housed or mounted straightforwardly in co,.ve..tionally populated printed-circuit boards 63.
The output stage of the evaluating means and at least the input stage of the access-c:ontrol block 30 ~or that entire block, as illus-trated) are fabricate~d as a unitary article, preferably a unitary integrated-circuit chip 62', on a single, preferably very small printed-circuit boarcl 62 (Fig. 21). This board 62 of Fig. 21 (to-gether with its chip 62') constitutes the unitary processor/actuator module 62 of Fig. 20.
~ 35 The evaluating ~eans "go / no-go" output 28 is advantageously buried deep within tha multiple layers of the chip 62'. Perhaps that vulnerabla output poin' 28 is accessible to high-technology reverse engineering that strips away and reads each chip layer in turn. That analytical mode, however, destroys the chip used for the analysis.
The resulting information cannot be used to preserve and restore operabilitv of the individual chip which is thus ~;es~cted, but only for use in capturing the design - preparatory to fabricating identi-CA 022ll9l0 l997-07-30 WO 96/24283 PCrlUS96/01615 6~
cal chips. Even a later-fabricated identical chip is of little use to the would-be misappropriator, particularly if it is completely identical, and the reference minutiae storage 25 is included within it but only part of the operating firmware is present on it.
A full-blown and very difficult circuit analysis, based on the dissection process, is required to det~rmin~ what parts of the device represent the reference minutiae, and how to change them. Det~ ~n-ing how to bypass the ~ _-rison process may be even more difficult.
To ~nh~nc~ the effectiveness of these obstacles to tampering and 10 reverse engineering, in principle it would be ideal to manufacture the chip 62' with a very distinctive and compound shape. Failing this the printed-circuit board 62 is made with such a shape.
The point here is to employ simple m~rh~nical means to place even further difficulties in the way of preparing a substitute device that can be dropped into place in the apparatus for security-bypass-ing purposes. The exemplary circuit board 62 has a slot 66 that allows motion of a tab ext~nA~ from the trigger - so that the trigger can be pulled.
Tm~i ately adjacent to this slot is a rather fine horizontal 20 projection that carries a n~ss~ry electrical contact 65 for actuat-ing the bullet 44 in the firing ~h~r 43. Thus it is awkward, though of course possible, to fashion a substitute or dummy proces-sor/actuator module that clears the trigger tab but makes the neces-sary contact.
Still further, the circuit board 62 inte~. icates with the rest of the circuitry. Such _ ication preferably is not through or.e or two electrical ines, carrying simple "go / no-go" signals in one direction, but rather a bus 68 of multiple finely and irregularly cp~c~ contacts that engage wiping contact fingers on the mating 30 board 63 and carry a complex of elaborately modulated signals in both directions. Such a contact structure is physically difficult to make, and even more difficult to analyze to det~ ne which lines may carry critical information or instructions.
Signals passing bidirectionally and concl~rrently across this interface enable the processor/actuator 62 to operate only if the mating board is connected, and conversely - somewhat deterring an operational, run-time analysis of circuit operation. Some of these signals may be ~ctually part of the fingerprint-da~a evaluating functions of the system, as the interface 68 of Fig. 21 corr~spon~c 40 to the ~h~ dogleg line passing through the evaluatiny -n~ block 27 in Fig. 20.

W096/24283 PCTnUS96/01615 -6s-Other dummy or pl A~O signals may be included simply to con~irm at both sides of the inter~ace that the contacts continue to be connected, or to confuse wo~ld-be analysts, or both. Em.ulation of such operation by a would-be tam.~erer is e~L ~ -ly difficult if not practically impossible. In event either side o~ the circuit fails to receive suitable acknowle~ f nt signals from the other side, the side r~o~n;zing such failure of acknowledy_.~..t automatically shuts down, or locks itself ~in~t further operation without a key input, or takes othe- tam.~er-defeating action appropriate to the particular 1o acknowlf ~f- - t that is absent.
Further ~althouqn slight) deterrence may be provided by recess-ing the receiving contacts within the mating board 63, so that attachment of probes while the boards are interconnected is diffi-cult; and by a Sf -w1lat ~l~horate com.~ound shape 69 etc. (Fig. 21).
Some additional active circuitry (not shown) is preferably positioned on the trigger tab, - icating with the pro~csor/act ator 62 through an additional set of wiping contacts 67 r~esse~
within the board, and mating contacts on the trigger tab. This arrangement enables cperation of the circuit but only when the trigger is pulled.
Preferably such provisions are in two stages, one set of con-tacts 67 connected so that detonation can occur only when the trigger is fully pulled, and another set (not shown) ~o~n~-~ted so that the fingerprint-evaluating elements of the apparatus work when the trigger is partly pulled. Alternatively for the latter function a separate power-on switch (not shown) in the handle may be actuated by simply grasping the handle.
The processor/actuato- ...od~le 62 of Fig. 20 thus make the measures required to breach the security system. at the evaluation output 28 lln~on~mic, at least unless the necessary effort, education and resources ~e~ are jusiified by ~cc~sc to a large n~mh~r of m;5~ppropriated w~po~C.
Other options include temporary disabling of the unitary pro-cessor/actuator ~odule 62 if an unauthorized person attempts to open ~ 35 the battery ~omr~rtme!nt cover 46c (Fig. 16) or other working-parts _ _-rtment of the we!apon. For this purpose advantageously a sensor switch 46s is mountedl directly to one of the circuit boards 63, to provide the circuit with information about the! ~r~rtment-cover 46c position. In designing such an arr~n~ --t it is ne~oCs~y to deal 40 with the possibility that the batteries 64 may be ~h~l~cted.
Such a dead-battery condition renders the circuit in~r~hle of tes ,ing whether the person opening the battery _ -rtment 46c is W096/2~283 PCT~S96/01615 authorized to do so. This consideration may suggest p-oy ~---ing the system to hold the c~m~rtment door 46c closed, in the absence of an authorized fingerprint, as lonq as power is available - e. a., with a battery-energized actuator latch 46a that is "normally open"
~unlatched if there is no power).
Such a system is fail safe in the particular sense that it prevents loss of use of the weapon by an owner; it permits anyone to open the _ -~tment 46c if the batteries 64 are dead. This operat-ing mode, however, would enable the tamperer to prevail merely by setting the system aside for a period of time sufficient to be sure that the batteries are dead.
A better approach, though perhaps slightly more costly, may be to mechanically keep the c _--tment locked if there is no battery power - by means of an alternative actuator latch 46a that is "normally closed" (latched if there is no power). This system instead holds the ~ _-~tment cover closed, in the ahs~nce of an authorized fingerprint, unless power is available.
In this case it is n~c~cs~ry to provide an auxiliary-power jack or connection point (not shown) - for supply of temporary power to operate the fingerprint-testing and lock-release circu-ts, just until the cnm~-tment can be opened and fresh batteries installed. The auxiliary jack must be suitably guarded Ag~i nct application of damaging high voltages, so that they do not result in op~ni ng of the c~r~tment cover 46c.
(c) Complementarv-detonator projectile - Physically unitary construction may not be an available tactic at the detonation point 31 (or, more generally for other types of energy-discharging means, the ~cc~ss-control output 31). Nevertheless functional inte de~en-30 dence can be employed to i~r~ unauthorized detonation.
The general principle of this strategy has already been de-scribed in the prece~i ng section, in regard to modulated signals p~csi ng bidirectionally across the interfaces 67, 68 of the proces-sor/actuator board 62. In the case now under discussion, the 35 pertinent interface 65 (Fig. 21) is between that circuit board 62 and a detonator circuit 49 (Figs. 19 and 22) within the c~ci ng of the bullet 44, via deton~tion contacts 43' that p~ss tnrough the wall 43 of the firing ~h~mh~r and mating contacts 49' that pass through the wall of the shell.
The detonator circuit 49 is preferably a very simple micropro-cessor and proy~ .~ble read-only memory ~"PROM") unit, which can hold certain codes and perform certain functions outlined below.

W096/24283 PCTrUS96/01615 To maintain re~o~le ~con~my in the manufacture of bullets 44, 44', the nl~r of contacts passing through the shell casing is preferably held to a rninimll~ Hence the opportunities for testing of circuit integrity at both sides of the interface are more restricted.
Nevertheless the detonation signals to the bullets can be made reAco~hly difficult 1;o emulate - at least within a short time after a weapon has been mi s-~rpropriated. One way to help in this effort is to make each bullet responsive to just one of a moderately large ~l~r (for example, t:hirty-two) of control codes, and to make each 10 weapon supply only one! such code to the detonator.
The authorized user may select one of the codes for entry into the storag~ memory 25 o~ the weapon, p~h~ps together with reference-minutiae data, and should personally memorize the code - but should not mark this information onto the weapon.
Corr~spon~i n~ly an authorized user can be certain to obtain bullets usable in that user~s weapon, but can then CQ~C~l from a would-be misappropriator of the weapon and bullets what code is needed to fire the bullets. For instance the control code for each p~ g~ of such bullets may be marked on a disposable outside wrap-20 ping of the p~k~g~, but not on the bullets themselves.
Alternatively the bullets may be sold in general form, and subject to code entry after purchase by use o- an electronic appli-ance that writes to the PROM in each detonator. The p~oy~ - ng appli~c~ preferably requires a user to m~n~-~ lly supply the desired code each time the appliance is used, so that theft of the appliance from a user's home or ~ffice does not reveal the code which resides in the memorv of that user's weapon.
In either event, the ~o~ing system makes the likelihood of z misappropriator's happl~ning to use the correct code rather small -though in the system described so far it can eventually be found bytesting, on average, a dozen or two differently coded bullets in the weapon. Ac will be appreciated such efforts at least are unlikely to be completed (and a co:rrectly st~bstitute-coded security-bypassing module 62 installed in the weapon) during, for example, the course of 35 an initial scuffle ove~ the weapon - or during the course of its theft from a home.
In .his scenario tne authorized user's module 62 prevents the mis~rpropriator from operating the weapon with bullets that are al-ready in it; the unknown code that is in the bullets deters prompt substitution and use oi- a bypass module. Nevertheless at least one added layer of securit~ is a~vant~g~otlcly included in the system.

_ _ _ _ _ _ _ _ _ _ _ _ =

W096/24283 PCT~S96/0161S

One such provision is that before supplying a detonation signal the weapon circuitry interrogates the detonator for its code. In event the detonation code ret~lrn~A is not correct, the weapon circuit 62 or preferably 63 disables itself by writing a disabling notation 5 to a nonvolatile memory element.
If desired, the system may be p-oy~~ ~' so that the weapon can be reenabled by writing a converse code into that memory element. To perform such reenabling, the p.o~.~. i ng should require (l) a factory technician with secret ~c~cs data, and/or (2) the physical presence 10 of an authorized person whose t '~rint is stored in the weapon.
Further complexity may be introduced through, for example, variable encryption of the detonator code.
Such systems deny the misappropriator of the weapon more than one single ~h~nre to guess what code is in use, before the weapon 1S becomes ~n~s~hle. Similarly - but more expensively - each bullet detonator circuit 49 can be p~oy -~ to p~-n~ntly ~isable itself upon receiving an incorrect detonation code, so that misappropriated -nition too becomes lln1lc~hle.
Provisions that make individual bullets relatively expensive are 20 not n~C~e~ily adverse to suco~ssful c ~rcial practice of the present invention. Most legitimate private users do not often fire large m~mh~rs of bullets; and in the professional context, as for example with law-enfo~ nt personnel, the improved security should be A? -~ adequate justification fo~ target practice with relatively 25 ~p~ncive bullets. In any event the self-disabling bullets may be offered as an option at slightly greater cost.
Another provision for an added layer of security is to incorpo-rate into the weapon the n~c~es~ry added circuitry and m-n~ con-trols to function as the code-writing appliance mentioned earlier in this section - but onlv when operated by tne authorized user.
Furth~ -re, the code that is written into each detonator PROM is not merely, _. q., an arbitrary five-bit code (one or thirty-two possible combinaticns) as before.
Rather it includes ar. encrypted version of a portion of the stored reference minutiae data 25. With _his refin~m~nt, the bullets as well as the weapon are completely personalized to the user.

At any rate ~he several antitampering features dis~lss~A in this section serve to provide the functional integration 64 of weapon circuitry and bullet circuitry. Those skilled in the art will r~ognize that the details of the invention as disclosed in this W096/24283 PCTnUS96/01615 section are straightforwardly adapted to utilization means 40 and energy-discharging means 44 (Fig. 18) of types other than bullets.

(d) More-severe self-disablinq features - As pointed out 5 above, the unitary actuator/processor module advan~g~o~cly has the capability of disabling itself by writing a suitable code to a p~-n~nt memory element. This arr~n~ - t may be subject to reenabling as described a~ove.
P1GY~ ~g the system in such ways may be seen as the least severe in a spectrum of measures that may be taken to deny use of the weapon to an unauthorized person. They are moderate measures by two criteria: they preserve the utility of the weapon for the authorized user, and do no harm to the would-be tamperer.
Such arr~n~_ ~t:s accordingly may be the most appropriate for a consumer whose major conc~rns are, for example, (l) attempts by a child to fire the weapon or take it apart to see what is inside, and (2) the possibility of being injured with the consumer's own weapon, in event of confronting a burglar or robber who wrestles the weapon from the owner. In this case the design philosophy is simply to 20 avoid injury to the child, the owner, and anyone else who may be nearby - as long as 1:he owner is not holding the weapon.
If the robber or burglar takes the weapon away, or the weapon is otherwise stolen, even though the thief may eventually find some way to reenable the weapon the consumer's primary objectives are met.
2S Such reenabling will be of auestionab e e~onomics from the thief's point of view (tQn~i ng to deter a practical thief from carrying the weapon away); and almc>st surely will occur 5~ ~~ 1.ere far from the site of the theft, so that injury to the owner and the owner's ~amily are correspo~ingly mc~re-remote possibilities.
At the other end of the spectrum of severe measures are p-oy~., ming and physical ele~lents that p~r--n~ntly disable or destroy essen-tially the entire wea~,on, and - in the process - incur a small pos-sibiity o~ serious injury to a careless thief or would-be tamperer.
Deliberately p, G~ ng such an apparatus to inflict serious injury upon the thief is no part of the teachings of this ~o~ nt.
It may be acceptable, however, to provide an interior movable barrel segment 43b (Fig. l9~ thai is biased by a spring (not shown) to deploy 43d automatically and p~rm~n~ntly in front of the firing ' -~. When so deployed the se_ -rt 43b becomes a barrel block 43b' - shown in phantom line.

W096/24283 PCTnUS96/0161S

The system is p~oy, -~ to operate the block-actuator latch 43a through electrical cn~n~ctions (not shown) from the circuit boards 63. When so p-oy~ ~d, the system automatically initiates the de-ployment 43b if any unauthorized person attempts to open the battery 6 - _-rtment 46c or other working-parts ~r~tment of the weapon.
The exclusive purpose of this arrangement is to prevent unautho-rized use of the weapon. If, however, a tamperer somehow n~g~q to detonate a bullet 44 in the firing ~h- '~r after such deployment, the weapon may explode or otherwise injure the tamperer.
Therefore it is n~ss~ry to provide and carefully test out a -~h~nical desigr. that mi ni mi zes this possibility - but also to provide r~con~le warnings. It will be understood that the configu-ration illustrated very schematically in Fig. l9 is not such a design, but only provides a representative means of p~r~-~ntly ~s disabling the weapon.
It appears essential that the weapon at least carry clearly lettered on its exterior a plainly worded warning of this arrange-ment, and particularly of the possibility of injury if a bullet is subsequently fired ~g~inst the block. Also advisable in addition is a recorded audibly spoken warning, actuated in response to initial efforts to open such a - _-rtment.
Specialized effort to det~rmin~ the legality of such measures is a ~C~cs~ry preli mi n~ y step to engaging in any such manufacture. It is important to make clear in advertising and at point of sale that a 25 weapon incorporating these more-severe antitam.pering measures is for sale only to a consumer who is in a position, and who intends, to safeguard the weapon against misuse by children etc. - and that weapons with the less-severe measures are alternatively available.
In providing these m~i ml~m-severity measures, the philosophy is 30 basically to sacrifice the weapon once in unauthorized hands, so as to p~-~ntly deny all possible profit from the effort and thereby deter f~Lule misappropriation. The power source in the weapon is available for this purpose.

Int~rm~iate on the scale of severity are measures that perma-nently disable or destroy selected n~qc~ry parts of the weapon, and that may p~r~rs inflict very minor injury to the hands of a person trying to use the gun or open a c~m~tment. In this case too, 40 warnings are essential. Possible options for such measures include:

W096/24283 PCT~US96/01615 ~ applying an ove!rload current to fuse a circuit element (not shown) within the unitary processor/actuator module 62 (Figs. 20 and 21), so that this module ho~ ~~ p~r~-n~ntly ;nop~ative and a factory repl~ nt is required to reenable the weapon; and ~ applying an overload ~u~-e,~t to fuse one or both of the detona-tion ~onn~ctions 43' (Fig. 19) through the firing ~h~mh~-~barrel wall 43, so that even "hot wiring" the contacts outside the ~ cr with a someho~q ~ellerated correct ignition code cannot apply a detonation signal to the projectile detonator.

Here too it is import:ant to make clear to potential buyers just what level of antitampering provisions are built into the weapon, and what alternatives are available.

5 . S IGNAL PROCESSII!~G

(a) Use of minutiae - The term "minutiaen means small details of a fingerprint, including ~n~in~s and bifurcations, islands, spurs and so on. The term ~co~r~ss~e abstract forms, e. q., coordinates that have been converted or transformed so as to describe the posi-tions of the physical objects relative to one another, or relative to some central or otherwise standardized feature.
2~ Here the tenm "feature" means a relatively large structure ~loop, whorl, axis, r~dges etc.), used to define a coordinate system in which to interpret the minutiae. Such features can be perceived and manipulated almost i..d~er.dently of coordinate system. Classi-cally, then, the abstract form of the minutiae is constructed by -30 reference to such topological near-invariants 2s the nl~mh~ of ridges between the physical minutiae.

(b) PreProcessinq to develop a direction maP - Each of the real-time print imagers described above has some fo~m Oc detector 7 (Figs. 23 and 1), which provides the FTIR-obt~in~ fingerprint data 21 to the new-minutiae deriving means 23. These data ~1 are rela-tively ragged and noisy.
Within the new-minutiae deriving means 23, preprocessing means 70 perform the first step in cl~ni ng up the daia: extraction of a direction map 71 (Figs. 24 and 23). This term refers to a systematic subdivision of the dat:a into regions or so-called "subpatches", for example rectangular or (for simplicity) preferably square pixel W096/24283 PCT~S96101615 groupings 80 analogous to the individual ~h~e or squares in a .-h~,~k~rho~ ~d.
Associated with each square 80 is a direction - e. a. horizon-tal 81, vertical 82, and all possible int~m~i~te orientations 83 -5 representing the average direction of the fingerprint ridge-and-y o~v~ structure within that square respectively. Such a map is then used to quide the orientation of a matched filter 72 for obt~ini ng refined information 73, 75, 24, 28 about the fingerprint ridges.
A direction map 71 has further usef~lln~cs in that it enables 10 determin~tion of the fingerprint axis. This information optionally (as suggested by the ~ch~ lines 71 in Fis. 23), but preferably, can be used in later stages 76, 27 of the processing to alleviate a major difficulty in print r~cognition, due to the plasticity of the skin over the bone of the finger.
For construction 70 of a direction map, four methods have been considered, of which three work about equally well. The methods are:
FFT detection, mean-gradient, and the projective Radon transform -and the less-successful method, the Hough transform.

(i) FFT detection was found ineffective because very few angles can be observed with small subpatches: the angular sensitiv-ity with a subpatch size of sixteen, for example, is about 30~.

(ii) Gradient method - In this method, the gradient of 25 each subpatch area 80 is det~ n~ by calculating small differences or differentials ~sx, ~ sy between values of the intensity-signal S, in two directions x and y respectively - _ being the horizontal direction (column number) and y the vertical (row number). For pixels i,i in generalized column _ and row i, and correspo~ingly 30 generalized nearby pixels ~-l,i (offset one column leftward) and i,i-1 (offset one row upward), the differences are defined by:

~ S x!i,j = S i,j S i,j-l and ~ sy~ =

Next these '?rs are used to obtain for the desired gradient a magnitude W = (~ S 2 + 1~ S 2) 1/2 40 and a direction. The direction is found by taking the function arctan(~ sy/~ sx) of the differentials.

W096/24283 PCTrUS96/01615 A ridge has two sides, and the angles of the gradients of these sides are in opposite directions. It is also desired to obtain the average angle. As the direction of such a ridge, or line, can only be defined over a span of O to 180C, an indirect t~chnique can be 5 used - defining the gradient as a complex number:

q~ exp [2l ~ arctan(Q sy/~ sx)].

The factor of two si~ply ~ELpS the larger angles, which are 180~ from the smaller angles on the opposite side of the ridge, onto the smaller values. The average angle is found by averaging the co~plex values and dividing by two, thus 6ri~e = ~ ~ ~ ~-arctan{imag[(~ ) avg. ] / realt(al,j) avg. ] }
where the functions ":imagn and "real" are respectively the imaginary and real parts of the averaged a~y~ -~t (~ij) avg; and the ~ ~ term is introduced ~-c~l~c~- the gradients are at right angles to the actual ridge lines.
In general this works better than the other direction-fi~; n~
algorithms, and it works for very small subpatches, where the FFT
method loses sensitivity. It fails in areas of the prints where all of the other methods f'ail - where noise from light contact, sweat gl ~n~c, and cuts make it hard to decide on a direction even by eye.
~iii) Radon Transform - This is a built-in function of the ~ -~cial MatLab~ analytical-mathematics software p~ e. It is related to the FFT approach and has similar limitations.

(c) Applvina the direction map to orient a tuned filter -After obt~i ni ~g a direction map 71, a tuned f-lter 72 is applied to the print i~ELge 21. The filter is asymmetric: it consists of a half-cycle cosine rolloff in the direction of the ridges, and a zero-mean l~-period, symmetric filter in the normal direction.
In algebraic notation the values of this filter 72, when it is not rotated, are -F(x) = cos~(x - xcenter)/widthl, along the filter, where the variable "width" is one parameter that is set for the ~ilter, based upon expected print cnaracteristics, and W096/2~283 PCT~S96/01615 F(y)O cost~ ~y Y c~ter) /w ri~e) if ;Y Y center! < ~w ri~e or F(y) b = b cos~ (Y ~ Y center) /w ri~e) otherwise.

Here W ri~e is a preestimated value for the ridge width - and also 5 for the y- oove width, which in this method is assumed to be the same as the ridge width.
The parameter b is evaluated in such a way as to make the integral over the area of the filter zero, after rotation (this to satisfy the condition that the filter be a "zero meann filter). The 10 parameter b is in other words evaluated so that after rotation:
W/2 W/2 width dxdy = ¦F~d_dy + JFbdx~y .
2 -width w/2 15 Experiment suggests that filter perform~nc~ would be better if the ridge and y~ OOve widths cou}d be adjusted in~p~n~ntly.
It is ideal to actually det~i nr- these variables for each authorized individual, and then set the weapon firmware accordingly.
It is not yet clear whether such individualized det~r~in~tion can be 20 automated for field conditions as might be desired (e. q., setting up the weapon in a dealer salesroom).
Customized analysis and setting would be extremely helpful, as the difference in results between data proc~cs~ with correct and incorrect direction maps (and therefore filter orientations) is 25 dramatic. When the direction is correct, all of the small trash in the image is gone, and the minutiae and features are clear.
When the direction is incorrect (usually 90c off due to pores and cuts, along with lighter pressure at tne edges of the contact area) the result is very messy, and fictional. In the absence of 30 techniques for reliably establishi~g the correct direction, some imp~v. -nt may be obt~i n~ through ~-ki ng the filter 72 longer.
An ll-by-ll matrix has been used to generate the rotated filter and tnen parse it down to a 9-by-9 neighborhood after rotation. This filter, set at one selected angle throughout each 16-by-16 or 32-by-35 32 block respectively, multiplies the brightness values 21 of thearea around each pixel by the correspon~i ng filter values. The sum of these products is -hen applied to the respective pixel.
This application of the mean angle over the full subblock or subpatch 80 is ~ e~uate in some areas, where the mean angle changes 40 drastically from block to block - as for example at a whorl in the fingerprint. It appears that in such regions an angular interpola-tion should be used from each block center to the adjacent block W096/24283 PCTrUS96/01615 center. This refjn~rr-~t wGuld also reduce the effects of gross direction errors.

(d) ~nh~n~jn~ binary character of the imaqe information -5 Since the previous fi.lter 72 (disussed ahove) is a ze~o -=n filter, reduction 74 of the i.mage to a binary one is straightforward: a threshold of zero can be applied. The result is a clear black-and-white representation.
Such a representation in turn can be r~tlc~ to lines by ~he 10 application of dila'ion and erosion algorithms, such as those found in the Matlab0 "Image Processing Toolbox". The overall product of the binarization, erosion and dilation is an output image 75 in which thin black lines ~p~r in place of the ridges.

(e) Abstractinq topoloqical near-invariants from the imaqe data - Once dilated, the iMage 75 is submitted to a further process-ing module 7~ for locating ~nutiae and identifying features - and then preferably creating an abstract of the minutiae, in which their locations relative to one another are expressed with reference to the 20 features (as distinguished from data-grid locations). The minutiae thus abstracted are collected in a tabulation 24 for use in a later evaluative c ~rison 27.
To locate the m~nutiae, the image 75 can be searched for ends by applying an end filte:r - which searches for a line ~n~i ng identified 25 as a connected pixel group that has a center pixel, one pixel in contact and no otners. If the same processing is repeated using an inverted binary image, the ridge bifurcations are found.
Each ridge island can then be found by identifying very short ridges - i. e., coun1:ing ridae pixels from end to end, and applying 30 an arbitrary length criterion - and each very short groove ("lake"
or "puddle") by simileLrly counting y~oov~ pixels.
Each such minutia found is pretabulated, categorized by type, as for example a location in the data grid. Next tne tabulation of locations is subjectecl to som~ counting scheme to find the n1~mn~r of ~ 3s ridges separating each relatively closely-spaced grouping (e. a., pair or triad) of minutiae, and the angle or angles at which the '~rs of the grouping are disposed relative to one another.
A gre~t variety of such counting s~~~~c may be devised for selecting the groupings and fi n~i ng the cpA~ings and angles. If there is no error, the ~llmh~r of ridges of separation should be in~p~n~nt of both the scheme employed and the finger orientation, twisting ~ue to grip approach etc., le~in~ to the image acqUisition.

W096/24283 PCT~S96101615 The angular data, however, typically will vary within some ranges, such as for instance ten to thirty degrees, that are charac-teristic of the apparatus and the subject person.

(f) Evaluatio~ ~i~ct near-invariant minutiae abstracted for the authorized person - The abstract minutiae 24 next proceed to evaluating means 27 for - -riSon ~in-ct analogous data 26, from reference-indicia storage 25, for the person or people authorized to use the weapon ~or other device). In this part of the process it is 10 essential to make proper allowance for the abovc ...~ntioned range of variation in angular data, or other variable data; in ~-ki ng this allowance it is helpful to use an actual range of variation measured for the authorized person.
All the above-described processes work very well as long as -~5 once again - there is no mistake in the direction of the ridges.
When there is an error in even one block, spurious ridge ~n~i n~S pop up all around the area of that block.
Such an ~rp~ance is a telltale signal of error to a trained human observer, but might be hard to train an automatic system to 20 discern. ~s will be seen in the following discussion, nonrandom pattern noise creates these errors; a strategy based on the expected, authorized person's fingerprint pattern would be most effective in overriding them. When this approach is taken, at least that person's print should be optimally recorded anà readily recognized - allowing identity verification with a high level of certainty for that person.
Another person's fingerprint may not produce a truly readable or pro~cs~hle image, or a print readable with high certainty. Program-ming can - in effect - use this fact itself, the fact of high _ certainty, as a disaualifying parameter.
It now ~ppe~s that the direction map used should be the ideal direction map of the authorized user, not a map derived from the pre-sented print. If the direction map agrees eve vwhere, a good binary image will result. If not, as is most likely if the print is a wrong one, the result will have errors and rejection will be likelier.
(g) Acquirinq data for the authorized user or users - An authorized user's characteristic dat~ ideally would be laboratory acquired. This may have the drawback of different meaau~ r,t conditions than in operation of the weapon, but may also provide some important benefits.
For example, steps may be taken to ensure that the direction map 71 found for an authorized user is clear and essentially correct.

W096/24283 PCTnUS96/01615 Also the full direct~on map thus acquired and used may be made larger than the area of the presented print: doing so would allow for the possihility that an acquired print in one instance includes certain peripheral areas, ancl in another instance includes others.
The full direction map can be shifted about to coincide with, for example, a high-pressure-point sample of directions of the trial print. Then the fu13., ideal direction map can be used to guide the filter over the tria] input.
The shifting of the filter 72 may then be used if desired to 0 form the basis in the abstraction-and-identification module 76, both for starting the minutiae~feature associations and as a first step in establishing ~h~ ti.on of invariants, such as ridge counts or the like as desired.
Laboratory acquisition may also enable use of mea~u ~,lent ~5 te~hni ques to optimiz:e the filter 72 periodicity, and ratio of ridge to y~oo~e widths, ancl also p~-h~rs to set the analytical apparatus 72-76 within the weapon optim~lly to take into ~cco~l~t whether the user's skin is charac:teristically dry etc.
Nevertheless some benefits of simplicity and efficiency, as weli 20 as identicality of m~!asurement conditions, do accrue from using the weapon (or other utilization means 40) itself as a primary apparatus for acquiring ~he authorized user's print data. Some of the same special optimization-c; described in the pr~c~i ng paragraphs may yet be applied even when the weapon is so used.
If desired the weapon may be attached, for this purpose, by an electrical umbilicus to laboratory equipment (not shown) - for a more complete analysis of print characteristics. Results of this analysis may then be written into various parts 70, 72, 25 of the electronic memory in the weapon.
An alternative approach is to use the apparatus of the weapon as the exclusive means of acauiring n~ data, but perhaps in a more-protracted acal~isition mode that takes several diff~rent sets of data under different conditions (e. a., filter periodicities) and ~ es perhaps the prebinarization contrast, or the postprocessing ~ 35 repro~ ihility, to select besr settings.
Fig. 23 shows t~his alternative approach. Initially the unitary processor/actuator mc,dule of Figs. 20 and 21 is hardwired so that the abstracted minutiae 24 from the new-minutiae deriving means 23 flow not only to the evaluating means 27 but also into storage 25 for the reference indicia.
When a weapon is first sold, it is operated to collect data from whoever shall be authorized to use the weapon. Preferably data sets W096/24283 PCT~S96/01615 for as many as several different trials are collected in the reference-indicia memory 25.
The evaluating means 27 are then operated to c _-~e the several data sets, and store cgn~n~ information about the c~-r~risons. In 5 particular the evaluating means det~rmine both the variability between sets (a measure of the variance that should be accepted by the device during actual use later) and the best set.
It may be found that some minutiae groupings are repro~ ihle among a first two or three data sets but not others, and other 10 minutiae groupings are found to be repro~llcihle among two or three of the "other" sets but not the first two. Such a case may p~-h~ps correspond to a situation in which a user habitually uses either of plural different ways of gripping the handle 46 and thumbrest l, 61 of the weapon.
In this case the apparatus may be p~oy~ .,~d to store co~ c~
information about a hybrid of two or more such data sets, or to store set-priority data indicating which of the data sets are encountered more frequently or matched more definitely. Such data may be used later to create suboptions for mat~hing a presented print, or to select the order in which the apparatus brings up the different data sets to try for a match.
When adequate image data and abstracted data sets have been collected, a rusible link 77 in the weapon may be severed, to prevent later flow of new-minutiae data 24 into the reference-indicia storage 25 25. (This may be done for example by applying a correct external voltage 75 to ~onn~ction points 78 across the link.) Equivalently a disable notation may be written into a suitable operating part of the circuit memory or firmware, to block such later data flow.
The reference-indicia and evaluating-means modules 25, 27 then 30 do not accept or process, respectively, any new user data - but can only operate to ~o~r~re a would-be user's submitted fingerprint image against the reference indicia, for generation of a positive or negative actuating signal 28 to the access-control module 30.
As to the "disable notation" mentioned just above, if desired the apparatus may be proy~ ..~d to allow overriding ~and overwriting) of such a notation - as îor example if the weapon changes ownership.
The apparatus may be p oy- -~ for example to enter such an override mode ~l) only in the physical presence of an authorized user, or ~2) only by a qualified t~hnician having n~c~SS~ry access codes, or ~3) 40 both - or, pre~erably, (4) any one of these three choices at the option of the person acquiring the weapon.

W096/24283 PCTrUS96/01615 (h) Practical r~ tions, and preferences for refinement - In a bright-field system the grooves appear as white as the un-touched prism illumin,ation, but dark areas, ~1-5~ by the absorption of the contacting ridges, are complex, include many small white dots 5 and are subject to a large variability in the degree of darkness.
The ~k~55 level is not repeatable, even though the artifacts in the ridges are quite stationary. In addition a ~n~ing pattern, at right angles to the ridge pattern, is caused by the pore struc-ture. When the skin is particularly dry this transverse fibrollcn~qs 10 can be very strong, and can reduce ridge lines to mere ~n~ clumps.
This is observed most o~ten in older subjects.
A locally adapt:ive asymmetric filter can suppress the clumping characteristic that otherwise makes it i~roscible to det~rmi n~ the minutiae and features in the prints.
The foregoing d:isclosure is int~n~ to be merely exemplary, and not to limit the scope of the invention - which is to be det~Tmi n~
by re~erence to the ~p~n~ ~1~ i mc,

Claims

WHAT IS CLAIMED IS:

1. Apparatus for acquiring surface-relief data from a relieved surface such as a finger; said apparatus comprising:
prism means formed from optical fibers and including:

a first end, comprising terminations of the fibers for contact with such relieved surface, and a second end, comprising opposite terminations of the same or corresponding fibers, for passage of light traveling along the fibers from the first end;

means for projecting light across the fibers in a region where the prism means satisfy at least one physical condition for efficient nonducting transillumination based upon a relationship between indices of refraction in the prism means;
said light being for lighting the first-end terminations;
wherein a light fraction dependent on contact between such relieved surface and each illuminated first-end termination is ducted from that termination along its fiber; and electrooptical means for receiving at the second end each light fraction from the first end, and in response forming an electrical signal characteristic of such surface relief.

2. The apparatus of claim 1, wherein said at least one physical condition comprises:
a limitation on maximum permissible numerical aperture of the prism means; and fiber diameter that is substantially constant with respect to longitudinal position.

3. A bright-field FTIR apparatus according to claim 2, wherein:
said projecting means comprise means for projecting light to enter individual fibers, through their respective side walls, immediately adjacent to their respective terminations; and each termination at said first end is oriented, relative to said projected light and relative to a longitudinal direction of its fiber, so that said light at that termination is:

reflected at that termination into and along its fiber toward the electrooptical means, if such relieved surface is out of contact with that termination, and in large part scattered by such relieved surface out of the corresponding fiber, if such relieved surface is in contact with that termination.

4. The apparatus of claim 3, wherein:
the projecting means direct the light into the prism means at angle, relative to an axis of the prism means, greater than twice the critical angle for the fiber cores against air, the critical angle being defined as the minimum angle off-normal for total internal reflection.

5. The apparatus of claim 3, wherein:
the critical angle for the fiber cores against air is roughly forty degrees, the critical angle being defined as the minimum angle off-normal for total internal reflection;
the first end of the prism means lies at an inclination angle of roughly forty-five degrees to an axis of the prism means; and the projecting means direct said light into the prism means at an angle to the central axis of roughly ninety degrees.

6. A dark-field FTIR apparatus according to claim 2, wherein:
said projecting means comprise means for projecting light to enter individual fibers, through their respective side walls, immediately adjacent to their respective terminations at the first end; and each termination at said first end is oriented, relative to said projected light and relative to a longitudinal direction of its fiber, so that said light at that termination is:

reflected at that termination out of its fiber, if such relieved surface is out of contact with that termination, and fractionally scattered by such relieved surface into and along its said fiber toward the electrooptical means, if such relieved surface is in contact with that fiber termination.

7. The apparatus of claim 6, wherein:
the critical angle for the fiber cores against air is roughly forty degrees, the critical angle being defined as the minimum angle off-normal for total internal reflection;
the first end of the prism means lies at an inclination angle of roughly ninety degrees to an axis of the prism; and the light from the projecting means is directed, when it is within the prism means, at an angle greater than roughly forty-five degrees to said prism axis, but less than said first-end inclination angle.

8. The apparatus of claim 6, wherein:
light from the projecting means enters the prism means with an initial intensity, and some of this projected light crosses the entire prism means; and the prism-means numerical aperture is small enough that projected light which crosses the entire prism means has at least roughly one tenth of the initial intensity, at least in a region adjacent to said first end.

9. The apparatus of claim 6, wherein:
the prism means comprise a section having extramural-absorption material, said section being between (1) a region where the light crosses the fibers and (2) the electrooptical means.

10. The apparatus of claim 9, wherein:
the EMA-material section has a numerical aperture that is substantially the same as numerical aperture of said region where the light crosses the fibers.

11. A dark-field FTIR apparatus according to claim 2, wherein:
the prism means comprise at least two separately fabricated fiber-optic optical elements secured together along a partially reflecting interface:

a first one of the elements comprising a prism that includes said first end and that is crossed by the light from the projecting means, and a second one of the elements that includes said second end;

said projecting means comprise means for projecting light into said prism to enter individual fibers, through their respective side walls, immediately adjacent to the partially reflecting interface;
the partially reflecting interface redirects some of the light from the projecting means to pass along said individual fibers within said prism toward their respective terminations at the first end;
said redirected light reaching each termination at said first end is:

reflected by the termination out of said corresponding fiber, if such relieved surface is out of contact with that termination, and fractionally scattered by such relieved surface back into the first prism and along said corresponding fiber, if such relieved surface is in contact with that fiber termination; and light fractionally scattered back into and along said corresponding fiber passes fractionally through the partially reflecting interface and the second element to the electrooptical means.

12. The apparatus of claim 2, wherein:
the second end of the prism means is different in cross-section from the first end; and the fibers are tapered to different cross-sections at the second end from that at the first end;
whereby the first end of the prism means is sized for contact with such surface, and the second end is sized for contact with the detector array.

13. The apparatus of claim 2, wherein:
the second end of the prism means is smaller in cross-section than the first;
in one segment of the prism means the fibers are tapered, so that they have smaller cross-sections at the second end than at the first;
whereby the prism means demagnify the surface-relief scattering pattern for application to the electrooptical means;
and the projecting means direct light into a segment of the prism means other than said one segment.

14. The apparatus of claim 2, wherein:
light from the projecting means enters the prism means with an initial intensity, and some of this projected light crosses the entire prism means; and the prism-means numerical aperture is small enough, at least where the light crosses the fibers, that projected light which crosses the entire prism means has at least roughly one tenth of the initial intensity.

15. The apparatus of claim 2, wherein:
the distance across the entire prism means is in a range of roughly one-and-a-half to two centimeters;
the prism means, at least where the light crosses the fibers, have a numerical aperture that does not exceed a corresponding range of roughly 0.46 to 0.42, respectively.

16. The apparatus of claim 2, wherein:
light from the projecting means enters the prism means with an initial intensity, and some of this projected light crosses the entire prism means; and the prism-means numerical aperture is small enough, at least where the light crosses the fibers, that projected light which crosses the entire prism means has at least roughly three-eighths of the initial intensity.

17. The apparatus of claim 2, wherein:
the distance across the entire prism means is in a range of roughly one-and-a-half to two centimeters;
the prism means, at least where the light crosses the fibers, have a numerical aperture that does not exceed a corresponding range of roughly 0.38 to 0.35, respectively.

18. The apparatus of claim 2, wherein:
the prism means, at least where the light crosses the fibers, have a numerical aperture that does not exceed 0.42.

19. The apparatus of claim 1, wherein:
the prism means comprise at least two separately fabricated optical elements secured together along a partially reflecting interface:

a first one of the elements comprising a fiber-optic prism that includes said first end and that is crossed by the light from the projecting means, and a second one of the elements comprising a fiber-optic taper that includes said second end.

20. The apparatus of claim 20, wherein:
said fiber-optic taper comprises extramural-absorption material.

21. The apparatus of claim 20, wherein said fiber-optic taper:
comprises extramural-absorption material; and has a numerical aperture greater than one-half.

22. The apparatus of claim 20, wherein:
said fiber-optic taper comprises extramural-absorption material, and has a numerical aperture greater than one-half; and said prism has a numerical aperture that does not exceed one-half.

23. Apparatus according to claim 1, particularly for acquiring surface-relief data from a human subject's finger; said apparatus further comprising:
means for stabilizing such human subject's finger against the first end of the prism means;
said stabilizing means comprising a handgrip for firm grasping by such human subject's hand with a particular finger braced against the first end of the prism means by said firm grasping.

24. The apparatus of claim 24, particularly for use by such human subject who has a thumb; and wherein:
such particular finger is such subject's thumb.

25. The apparatus of claim 24, wherein:
said stabilizing means comprise means for orienting such finger on the first end of the prism means within a relatively narrow range of finger positions;
said handgrip comprises means for orienting such hand, in relation to the first end of the prism means, within a relatively narrow range of hand positions;
wherein orientation of such hand by said handgrip inherently constrains orientation of such finger, on the first end of the prism means, within said relatively narrow range of finger positions.

26. The apparatus of claim 1, particularly for use in testing whether said relieved surface is a particular specified relieved surface; and wherein:
the electrooptical means comprise an electrooptical detector array for receiving at the second end of the prism means said light fraction from each termination at the first end, and in response thereto providing a corresponding array of electrical signals which are characteristic of such surface relief; and said signal formed in response to each light fraction comprises the array of electrical signals;
the electrooptical means comprise means for comparison of said electrical-signal array with data for such particular specified relieved surface, to generate at least one comparison-result signal which is characteristic of the results of said comparison;
the comparison means comprise means for processing the electrical-signal array by application thereto of a direction map which is characteristic of such particular specified relieved surface; and said signal formed in response to each light fraction further comprises the comparison-result signal.

27. The apparatus of claim 27, wherein:
the processing means further comprise means for applying a tuned filter, oriented in accordance with the direction map, to the electrical-signal array to obtain a filtered image.

28. The apparatus of claim 27, wherein the processing means comprise:
means for applying a center and an axis of the direction map to determine coordinates of the relieved-surface details; and means for using the axis of the direction map to control the means for comparison.

29. The apparatus of claim 1, wherein:
the at least one condition or the prism means comprise, at least in a region where the light crosses the fibers, a numerical aperture that does not exceed 2navg (D/xF)1/4, where navg ~ average of core and cladding refractive indices in said region of the prism means;
D ~ periodicity of the fiber structure in said region; and XF ~ illumination-path distance in said region;

said illumination-path distance being measured across the prism means if the apparatus is illuminated through one said side face, and being measured to the midplane of the prism means if the apparatus is illuminated through more than one said side face.

30. The apparatus of claim 1, wherein:
said at least one condition of the prism means comprises, at least where the light crosses the fibers a numerical aperture that does not exceed one-half.

31. The apparatus of claim 1 wherein:
said at least one condition of the prism means comprises, at least in a region where the light crosses the fibers, a numerical aperture that does not exceed 2n avg (D/2xF)1/4, where navg ~ average of core and cladding refractive indices in said region of the prism means;
D ~ periodicity of the fiber structure in said region; and XF ~ illumination-path distance in said region;

said illumination-path distance being measured across the prism means if the apparatus is illuminated through one said side face, and being measured to the midplane of the prism means if the apparatus is illuminated through more than one said side face.

32. The apparatus of claim 1, wherein:
said at least one condition of the prism means comprises, at least where the light crosses the fibers, a numerical aperture that does not exceed 0.42.

33. Apparatus according to claim 1, in further combination with:
means for controlling access to facilities, equipment, a financial service or information;
said access-controlling means comprising means for processing the signal formed by the electrooptical means to check identity of the relieved surface, and for applying the results of the check to control access to such facilities, equipment, financial service or information.

34. Apparatus according to claim 1, in further combination with:
a secured system subject to access control based upon surface-relief data from a relieved surface such as a finger;
said system comprising utilization means, susceptible to misuse in the absence of a particular such relieved surface that is related to an authorized user, said utilization means being selected from the group consisting of:

a facility, apparatus, means for providing a financial service, and means for providing information; and means for processing the signal formed by the electrooptical means to check identity of the relieved surface, and for applying the results of the check identity to control access to the utilization means.

35. Apparatus according to claim 1, in further combination with:
a personal weapon subject to access control based upon surface-relief data from a human user's finger;
said weapon comprising means for discharging an energy-transmitting agency to influence an adversary, said agency-discharging means requiring enablement for its operation; and means for processing the signal formed by the electrooptical means, to check identity of the relieved surface, and for applying the results of the check to control enablement of the agency-discharging means.

36. The weapon or claim 35, further comprising:
a handgrip for firm grasping by such human subject's hand to support the weapon, with a particular finger braced against the first end of the prism means by said firm grasping.

37. The weapon of claim 35, wherein:
at least a part of said discharging means and at least a part of said electronic means are physically formed together as a substantially unitary electronic module to deter unauthorized bypassing of said fingerprint-data access control.

38. The weapon of claim 37, wherein:
the substantially unitary electronic module comprises a complicatedly contoured shape that matches and is required by an electronic-module receptacle of the weapon;
said shape including electrodes for contacting elements of the weapon to effect said discharging.

39. The weapon of claim 37, wherein:
said at least a part of said discharging means comprises means for providing a specifically controlled electrical impulse to effect said discharging;
the specifically controlled electrical impulse has a characteristic selected from the group consisting of:

duration within a relatively narrow range, voltage within a relatively narrow range, a particular waveform, a data stream conveying particular information, and any combination of two or more of the foregoing four characteristics; and the energy-discharging agency is manufactured to respond exclusively to said specifically controlled electrical impulse.

40. The weapon of claim 39, wherein:
the energy-discharging means comprise a bullet having therein a charge of explosive powder and an electrical detonator to ignite the explosive charge; and the detonator is manufactured to respond exclusively to said specifically controlled electrical impulse.

41. Apparatus for acquiring surface-relief data from a relieved surface such as a finger; said apparatus comprising:
prism means formed from optical fibers and including:

an optical data-input end, comprising terminations of fibers of the bundle, for contact with such relieved surface at the fiber terminations;

at least one side face for receiving light into the prism means in a region where indices of refraction of the prism means are sufficiently close to one another for efficient nonducting illumination, and a partially reflecting surface for redirecting light received through the side face to illuminate the fiber terminations at the data-input end; and means for receiving light that passes through the partially reflecting surface from the illuminated fiber terminations, and in response providing at least one signal which is characteristic of such surface relief.

42. Apparatus according to claim 41, further comprising:
means for projecting light into the light-receiving face of the prism so that the light crosses part of the bundle to enter individual fibers through their respective side walls immediately adjacent to the partially reflecting surface, to be partially redirected by said surface toward the data-input end of the prism for illuminating the terminations at the data-input end of the prism.

46. The weapon of claim 45, wherein:
the substantially unitary electronic module comprises a complicatedly contoured shape that matches and is required by an electronic-module receptacle of the weapon;
said shape including electrodes for contacting elements of the weapon to effect said discharging.

47. The weapon of claim 44, wherein:
the developing-and-applying means comprise a crosslit optic-fiber prism with numerical aperture not exceeding one-half, for collecting fingerprint data by frustrated total internal reflection.

48. The weapon of claim 44, wherein said bypass-deterring means comprise:
means, associated with said discharging means, for providing a specifically controlled electrical impulse to actuate said energy-transmitting agency;
said specifically controlled electrical impulse having a characteristic selected from the group consisting of:

duration is within a relatively narrow range, voltage is within a relatively narrow range, a particular waveform, a data stream conveying particular information, and any combination of two or more of the foregoing four characteristics.

49. The weapon of claim 48, wherein:
the energy-transmitting agency is a bullet having therein a charge of explosive powder and an electrical detonator to ignite the explosive charge; and the detonator is manufactured to respond exclusively to said specifically controlled electrical impulse.

50. The weapon of claim 44, wherein the bypass-deterring means comprise:
means for providing entry to working parts of the weapon;
second means for applying the electronic signal to control enablement of the entry-providing means;
means for determining when entry to working parts of the weapon is gained without operation of the second means;
means responsive to the entry-determining means for substantially permanently disabling the weapon when entry to working parts of the weapon is gained without operation of the second means.