|Publication number||USRE42070 E1|
|Application number||US 12/401,461|
|Publication date||Jan 25, 2011|
|Filing date||Mar 10, 2009|
|Priority date||Mar 27, 2000|
|Also published as||US7103223, US20030219158|
|Publication number||12401461, 401461, US RE42070 E1, US RE42070E1, US-E1-RE42070, USRE42070 E1, USRE42070E1|
|Original Assignee||Look Dynamics, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (15), Non-Patent Citations (3), Classifications (18), Legal Events (3)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This patent application is a continuation-in-part of U.S. patent application Ser. No. 10/051,364, filed in the U.S. Patent and Trademark Office on Jan. 18, 2002 now abandoned, which is a continuation in-part of U.S. patent application Ser. No. 09/536,426, filed in the U.S. Patent and Trademark Office on Mar. 27, 2000 now U.S. Pat. No. 6,678,411.
1. Field of the Invention
This invention relates generally to spatial light modulators and, more particularly, to a spatial light modulator with radially oriented active light modulating sectors for radial and angular analysis of beams of light, including Fourier transform optic patterns, for uses such as characterizing, searching, matching or identifying shape content of images.
2. State of the Prior Art
There are situations in which useful information can be derived from spatially dispersed portions of light beams. In particular, when an image is being carried or propagated by a light beam, it may be useful to gather and use or analyze information from a particular portion of the image, such as from a particular portion of a cross-section of a beam that is carrying an image.
For example, in my co-pending U.S. patent application, Ser. No. 09/536,426, which is incorporated herein by reference, narrow, radially oriented portions of a Fourier transform of an image are captured and detected in the spatial domain and used to characterize and encode images by shape for storage, searching, and retrieval. As explained therein, such radially oriented, angularly or rationally spaced portions of light energy from a Fourier transform, i.e., Fourier domain, of an image are captured sequentially in the spatial domain by positioning a rotating, opaque mask or wheel with a radially oriented slit in the Fourier transform domain of a light beam carrying the image after passing the light beam through a Fourier transform lens and detecting the light in the spatial domain that passes through the slit at various angular orientations, i.e., degrees of rotation. The light energy detected in the spatial domain that is passed through the slit in the Fourier domain at each angular orientation is characteristic of the portions of the image content that are generally linearly aligned in the same angular orientation as the slit in the rotating mask when the light energy is detected.
That system with the rotating, radially oriented, slit does perform the task of characterizing and encoding images by shape content of the images quite well and quite efficiently. However, it still has several shortcomings. For example, resolution of spatial frequency of an image at each angular orientation of the rotating slit is not as good as some applications or uses of such a system might require. Also, the spinning mask or wheel with an associated drive mechanism, like all mechanical devices, has stability and long term reliability issues, not to mention size and weight requirements.
Accordingly, it is a general object of this invention to provide an improved apparatus and method for capturing and recording optical information from portions of optical images.
A more specific object of this invention is to provide an improved apparatus and method for spatial analysis of Fourier transform optical patterns of images for shape content of such images.
Another specific object of this invention is to provide an improved apparatus and method for characterizing and encoding images by shape content for storing, searching, comparing, matching, or identifying images.
This and other objects, advantages, and novel features of the invention shall be set forth in part in the description that follows, and in part will become apparent to those skilled in the art upon examination of the following description or may be learned by the practice of the invention. The objects and the advantages may be realized and attained by means of the instrumentalities and in combinations particularly pointed out in the appended claims.
To further achieve the foregoing objects, the apparatus of this invention includes a spatial light modulator with a plurality of addressable, active optic elements that extend radially at various angular orientations in relation to an axis. The active optic elements are preferably shaped to modulate portions of light beams incident on discrete sectors of an active optic area on which the beam of light can be focused. Therefore, active optic modulators in the shape of individual sectors, i.e., essentially wedge-shaped, are preferred, although other shapes are also feasible and, in special circumstances, possibly even more desirable, such as rectangular for better resolution or curved for detection of curved shape content of an image. For better resolution of spatial frequency of shape content, the radially extending wedges or rectangles of the active optic area can be comprised of individually addressable segments, which can be activated separately or in groups, depending on the resolution desires. Wedge-shaped sectors can comprise segments of smaller, truncated wedge-shaped active optic elements or groups of sensors in pixel arrays that, in composite, form such shapes. Rectangular areas can also comprise smaller rectangular segments or composited groups of sensors in pixel arrays to form such radially extending, angularly spaced, active optic components or areas. For shape content characterization of an image, an optic pattern that is a Fourier transform of the image is focused on the active optic area, and radially disposed portions of the Fourier transform optic pattern at various angular orientations are selected and isolated by the spatial light modulator and projected into the spatial domain for detection of shape content of the image that is aligned with such angular orientations. The intensities of light detected from such respective portions are characteristic of such shape content, including brightness, sharpness, orientation, and density or spatial frequency of image features, and can be recorded, stored, or used to compare with similarly analyzed shape content of other images to find and identify matches or near matches of images with such shape content. Optional image pre-processing to add ghost images at various radial and angular relationships to the image and at various light intensities can enhance detectability of shape content and can enable near matching of images with similar shape content.
The accompanying drawings, which are incorporated in and form a part of the specification, illustrate the preferred embodiments of the present invention, and together with the descriptions serve to explain the principles of the invention.
In the Drawings:
FIGS 10a-c include diagrammatic, elevation view of the active light modulating components of the segmented radial spatial light modulator device to illustrate a use of an outer segment of a vertically oriented sector of the light modulation components of the segmented radial spatial light modulator device of this invention along with diagrammatic views of an image being characterized and a resulting detectable optic pattern that is characteristic of some of the vertically oriented shape content of the image;
A segmented radial spatial light modulator (SLM) device 50 according to the present invention is illustrated diagrammatically in
An enlarged elevation view of the active optic area 54 of the integrated circuit 52 is illustrated in
The selection and isolation of a portion of the incident light beam 27(p) is illustrated in
The system 10 for characterizing, encoding, and storing images by shape content of such images, as illustrated diagrammatically in
The images 12, 14, . . . , n can be in virtually any form, for example, visual images on photographs, films, drawings, graphics, arbitrary patterns, ordered patterns, or the like. They can also be stored and/or generated in or from digital formats or analog formats. Such images can have content that is meaningful in some manner when viewed by humans, or they can appear to be meaningless or not capable of being interrupted by humans but characteristic of some other content, e.g., music, sounds, text, software, and the like. Essentially, any optic pattern of light energy intensities that can be manifested or displayed with discernable shape content can be characterized and encoded with this system 10.
A sample image 12, which can be obtained from any source (e.g., Internet, electronic data base, web site, library, scanner, photograph, film strip, radar image, electronic still or moving video camera, and other sources) is entered into the optical image shape characterizer 10, as will be described in more detail below. Any number n of other sample images 14, . . . , n, are shown in
In the embodiment of the system 10 illustrated in
As mentioned above, the image 12 can be entered into the optical image characterizer 10 by the computer 20 and E-SLM 26, as will be described in more detail below. However, the image 12 will undergo a significant transformation upon passing through the thin, positive lens 30, also called the Fourier transform (FT) lens. A Fourier transform (FT) of the sample image 12′ rearranges the light energy of the optic pattern of image 12′ into a Fourier transform (FT) optic pattern 32, which is unique to the image 12′, even though it is not recognizable as the image 12′ to the human eye and brain, and which can be characterized by intensities, i.e., amplitudes, of light energy distributed spatially across the optic pattern 32. The complex amplitude distribution of light energy 34 in the optic pattern 32 is the Fourier transform of the complex light distribution in the image 12′. Image 12′ is a recreation of the image 12 in monochromatic, preferably coherent, light energy, as will be described in more detail below, although white light will also work. Of course, persons skilled in the art will also recognize that an E-SLM is only one of a number of well-known devices that can create the image 12′ in monochromatic or white light, and this invention is not limited to this particular example.
Concentrations of intense light energy in the Fourier transform (FT) optic pattern 32 generally correspond to spatial frequencies of the image 12′, i.e., how closely together or far apart features in the image 12′ change or remain the same. In other words, spatial frequencies are also manifested by how closely together or far apart light energy intensities across the light beam 27 change or remain the same. For example, a shirt with a plaid fabric in an image (not shown), i.e., having many small squares, would have a higher spatial frequency, i.e., changes per unit of distance, than a plain, single-color shirt (not shown) in the image. Likewise, portions of an image, such as the bumper and grill parts 35 of the automobile in image 12′, would have a higher spatial frequency than the side panel 36 portion of the automobile image 12′, because the bumper and grill parts 35 comprise many small pieces with various edges, curves, and other intricate changes within a small spatial distance, whereas the side panel 36 is fairly smooth and uniform over a large spatial distance. Light energy from the finer and sharper details of an image (more spatial frequency), such as the more intricate bumper and grill parts 35 of the image 12′, tend to be dispersed farther radially outward from the optical center or axis 40 in the Fourier transformed image than light energy from more course or plain details of an image (less spatial frequency), such as the side panel 36 of the image 12′. The amplitude of light energy 34 dispersed radially outward in the Fourier transform optic pattern 32 is related to the light energy of the corresponding portions of the optic pattern of image 12′ from which such light energy emanates, except that such light energy is concentrated into areas or bands 34 at the plane of the Fourier transform (FT) optic pattern 32 after they are refracted by the FT lens 30, i.e., into bands of intense light energy separated by bands of little or no light energy, which result from constructive and destructive interference of the diffracted light energy. If the high spatial frequency portions of the image 12′, such as the bumper and grill portion 35, are bright, then the intensity or amplitude of light energy from those high spatial frequency portions of the image 12′, which are dispersed by the FT lens 30 to the more radially outward bands of light energy 34 in the Fourier transform optic pattern 32, will be higher, i.e., brighter. On the other hand, if the high spatial frequency portions of the optic pattern of image 12′ are dim, then the intensity or amplitude of light energy from those high spatial frequency portions of the optic pattern of image 12′, which are dispersed by the FT lens 30 to the more radially outward bands of light energy 34 in the Fourier transform optic pattern 32, will be lower, i.e., not so bright. Likewise, if the low spatial frequency portions of the optic pattern of image 12′, such as the side panel portion 36, are bright, then the intensity or amplitude of light energy from those low spatial frequency portions of the optic pattern of image 12′ which are dispersed by the FT lens to the less radially outward bands of light energy 34 in the Fourier transform optic pattern 32 (i.e., closer to the optical axis 40), will be higher, i.e., brighter. However, if the low spatial frequency portions of the optic pattern of image 12′ are dim, then the intensity or amplitude of light energy from those low spatial frequency portions of the optic pattern of image 12′, which are dispersed by the FT lens 30 to the less radially outward bands of light energy 34 in the Fourier transform optic pattern 32, will be lower, i.e., not so bright.
In summary, the Fourier transform optic pattern 32 of the light emanating from the image 12′: (i) is unique to the image 12′; (ii) comprises areas or bands of light energy 34 concentration, which are dispersed radially from the center or optical axis 40, that represent spatial frequencies, i.e., fineness of details, in the image 12′; (iii) the intensity or amplitudes of light energy 34 at each spatial frequency area or band in the Fourier transform optic pattern 32 corresponds to brightness or intensity of light energy emanating from the respective fine or course features of the image 12′; and (iv) such light energy 34 in the areas or bands of the Fourier transform optic pattern 32 are detectable in intensity and in spatial location by this invention.
Since this optical image characterizer 10 of this invention is designed to characterize an image 12 by shapes that comprise the image 12, additional spatial filtering of the Fourier transform light energy pattern 32 is used to detect and capture light energy emanating from the finer or sharper details or parts of such finer or sharper details in the image 12′, which are aligned linearly in various specific angular orientations. Such spatial filtering can be accomplished in any of a number of different ways, as will be explained in more detail below, but an exemplary spatial filter arrangement for this function is included in a combination of the segmented radial spatial light modulator device 50 with the polarizing beam splitter 70. Essentially, the segmented radial SLM device 50 rotates the plane of polarization of selected portions of the Fourier transform optic pattern 32 from p-plane polarization to s-plane polarization, or vice versa, as explained above, and the polarizing beam splitter 70 separates light energy of those portions that is isolated and polarized in one plane from the light energy of the rest of the Fourier transform optic pattern 32 that remains polarized in the other plane so that such light energy of the selected and isolated portions can be detected separately.
In the example, segmented radial SLM 50 shown in
Of course, segments in different sectors of different angular orientations about the optic axis 40 will align linearly with features or lines in the image 12′ that have different angular orientations, as will be described in more detail below. Thus, the light energy bands 62 in the filtered pattern 60 will change, as active optic segments in different sectors are selected and activated, to represent different features, details, edges, or lines in the optical pattern of image 12′ at various angular orientations, intricateness or fineness, and brightness, as will be explained in more detail below. In general, however, the light energy bands 62, if inverse Fourier transformed from the FT optic pattern 32 after the above-described spatial filtering 54, will be located in the same spatially-related sites as the features in the original image 12′ from which such light energy emanated. For example, light energy in a band 62 in pattern 60 that originally emanated from bumper and grill parts 35 in image 12′, after spatial filtering with the vertical sector of the bumper and grill parts 35 in image 12′.
The spatially filtered light energy in bands 62 of the filtered pattern 60 can be detected by a photodetector 80 at any of the various angular orientations of the activated sectors and fed electronically to a computer 20 or other microprocessor or computer for processing and encoding. While only one photodetector 80 with an example 16×16 array 82 of individual photosensitive energy transducers 84 is illustrated in FIG. 5 and is sufficient for many purposes of this invention, other detector arrangements, for example, the two offset detector arrays described in co-pending patent application, Ser. No. 09/536,426, or one or more larger detector arrays, could also be used.
The computer 20, with input of information about the filtered optical patterns 60, i.e., light energy intensity (I) distribution, from the detector array 82, along with information about the image 12 (e.g., identification number, source locator, and the like), information about the angular orientation (R) of the sector in which a segment is activated, and information about the radial distance or scale (S) of the activated segment relating to spatial frequency, can be programmed to encode the characteristics of the image 12 relating to the shape content of the image 12. One useful format for encoding such information is by pixel of the filtered image 60, including information regarding x, y coordinate location of each pixel, Rotation (i.e., angular orientation of the sector in which a segment is activated, thus of the linear features of the image 12 that align with such angular orientation), and Intensity (i.e., amplitude of light energy from the filtered pattern 60 that is detected at each pixel at the angular orientation R. A searchable flag, such as a distortion factor X, can also be provided, as explained in more detail co-pending patent application, Ser. No. 09/536,426 or by the ghost image pre-processing feature of this invention, which will be explained in more detail below. Such combination of angular orientation or rotation R, light energy intensity I for each pixel, and distortion factor X can be called a “RIXel” for short. Scale (i.e., spatial frequencies of image 12 content at such angular orientations) can also be included in such encoding, if desired. When including a scale factor S, the combination can be called a “RIXSel”. Each RIXel or RIXSel can then be associated with some identifier for the image 12 from which it was derived (e.g., a number, name, or the like), the source location of the image 12 (e.g., Internet URL, data base file, book title, owner of the image 12, and the like), and any other desired information about the image, such as format, resolution, color, texture, content description, search category, or the like. Some of such other information, such as color, texture, content description, and/or search category, can be information input from another data base, from human input, or even from another optical characterizer that automatically characterizes the same image 12 as to color, texture, or the like—whatever would be useful for searching, finding, or retrieving image 12 or for comparing image 12 to other images.
Some, all, or additional combinations of such information about each image 12, 14 . . . , n characterized for shape and encoded, as described above, can be sent by the computer 20 to one or more data base(s) 102. Several example data base architectures 104, 106, 108 for storing RIXel or RIXSel information about each image 12, 14, . . . , n are shown in
In the optical image characterizer 10 illustrated in
The beam 25(s) is then passed through a polarizing beam splitter 116, which reflects light polarized in one direction at plane 118 and transmits light polarized in the orthogonal direction. In this example, the polarizing beam splitter 116 reflects s-polarized light and transmits p-polarized light, and it is oriented to reflect the s-polarized beam 25(s) toward the electrically addressed spatial light modulator (E-SLM) 16. The monochromatic, preferably coherent, light beam 25(s) incident on the E-SLM 36 provides the light energy that is utilized to carry the shape content of the image 12′ for further analysis, characterization, and encoding according to the principles of this invention in the examples described below.
As mentioned above, there are many ways of “writing” images 12, 14, . . . , n into a light beam, one of which is with an E-SLM 16. In this example, computer 20 has the content of image 12 digitized, so the computer 20 can transmit digital signals via link 21 to the E-SLM 26 in a manner that addresses and activates certain pixels in the E-SLM 26 to “write” the image 12′ into reflected light beam 27(p), as is understood by persons skilled in the art. Essentially, the addressed pixels rotate the plane of polarization by 90 degrees from the s-plane of incident beam 25(s) to the p-plane of reflected beam 27(p), or by some lesser amount for gray-scales, in a manner such that the reflected light energy with partially or fully 90-degree polarization plane rotation is in a monochromatic optical pattern of the image 12′. Of course, persons skilled in the art will also understand that the image 12′ could also be created with an E-SLM that operates in an opposite manner, i.e., the plane of polarization is rotated in reflected light, except where pixels are activated, in which case the computer 20 would be programmed to activate pixels according to a negative of the image 12 in order to write the image 12′ into reflected beam 27. Either way, the emerging beam 27(p) of coherent light, carrying image 12′, is p-polarized instead of s-polarized or vice versa. Consequently, in the above example, the monochromatic light beam 27(p), with its light energy distributed in an optic pattern that forms the monochromatic image 12′, is transmitted by the polarizing beam splitter 116 to the FT lens 30, instead of being reflected by it.
The positive Fourier transform (FT) lens 30, as explained above is positioned in the light beam 27(p) and redistributes the monochromatic light energy from the image 12′ into its Fourier transform optic pattern 32, which occurs at the focal plane of the FT lens 30. Therefore, the segmented radial SLM 50 of this invention is shown positioned in the focal plane of the FT lens 30, as indicated by the focal distance F in
Of course, the Fourier transform optical pattern can also be projected and captured or spatially filtered according to this invention in locations other than the focal plane of the lens 30 by appropriate optical components or systems (not shown), as is well-known by persons skilled in the art. For example, but not for limitation, another lens (not shown) positioned behind the focal plane of the lens 30 can form and project an enlargement of the FT image in a different plane, as explained in the U.S. Pat. No. 3,771,124, issued to D. McMahon. Therefore, while the example implementation of this invention shown in
The Fourier transform optic pattern 32, as mentioned above, is symmetrical from top to bottom and from left to right, so that each semicircle of the Fourier transform optic pattern 32 contains exactly the same distribution and intensity of light energy as its opposite semicircle. Light energy from lower spatial frequencies in the image 12′ are distributed toward the center or optical axis 40 of the Fourier transform optic pattern 32, while the light energy from higher spatial frequencies in the image 12′ are distributed farther away from the optical axis 40 and toward the outer edge of the pattern 32, i.e., farther radially outward from the optic axis 40. Light energy from features in the image 12′ that are distributed vertically in the image 12′ to create those various spatial frequencies is likewise distributed vertically in the Fourier transform optic pattern 32. At the same time, light energy from features in the image 12′ that are distributed horizontally in the image 12′ to create those various spatial frequencies is distributed horizontally in the Fourier transform optic pattern 32. Therefore, in general, light energy from features in the image 12′ that are distributed in any angular orientation with respect to the optical axis 40 to create the various spatial frequencies in the image 12′ is also distributed at those same angular orientations in the Fourier transform optic pattern 32. Consequently, by detecting only light energy distributed at particular angular orientations with respect to the optical axis 40 in the Fourier transform optic pattern 32, such detections are characteristic of features or details in the image 12′ that are aligned linearly in such particular angular orientations. The radial distributions of such detected light energy at each such angular orientation indicate the intricateness or sharpness of such linear features or details in the image 12′, i.e., spatial frequency, while the intensities of such detected light energy indicate the brightness of such features or details in the image 12′.
Therefore, a composite of light energy detections at all angular orientations in the Fourier transform optic pattern 32 creates a composite record of the shapes, i.e., angular orientations, intricateness or sharpness, and brightness, of linear features that comprise the image 12′. However, for most practical needs, such as for encoding shape characteristics of images 12, 14, . . . , n for data base storing, searching, retrieval, comparison and matching to other images, and the like, it is not necessary to record such light energy detections for all angular orientations in the Fourier transform pattern 12′. It is usually sufficient to detect and record such light energy distributions and intensities for just some of the angular orientations in the Fourier transform optic pattern 32 to get enough shape characterization to be practically unique to each image 12, 14, . . . , n for data base storage, searching, and retrieval of such specific images 12, 14, . . . , n. For purposes of explanation, but not for limitation, use of 11.25-degree angular increments is convenient and practical, because there are sixteen (16) 11.25-degree increments in 180 degrees of rotation, which is sufficient characterization for most purposes and has data processing and data storage efficiencies, as explained in co-pending U.S. patent application Ser. No. 09/536,426. However, other discrete angular increments could also be used, including constant increments or varying increments. Of course, varying increments would require more computer capacity and more complex software to handle the data processing, storing, and searching functions.
In the preferred embodiment of this invention, the segmented radial SLM 50, shown in
The preferred, but not essential, shape of the active optic sectors, e.g., sector 500, in the segmented radial SLM 50 is a narrow, elongated wedge. The width of the wedge will depend on the light energy available or needed and the optic resolution desired. A wider sector will direct more light energy 34 to the detector 80, but precision of line or feature resolution of the image 12′ will degrade slightly. A narrower sector will get better line resolution, but with a corresponding increase in the complexity of the resulting pattern shape generalization and complexity and a decrease in light energy directed to the detector 80. There may also be a practical limitation as to how narrow and close the wedges can be made with the connecting electric traces in a limited active optic area 54 in an economic and efficient manner. Therefore, a desirable balance between these resolution, detectability, and size considerations may be struck in choosing sector size. Also, for specialized applications, sectors of different shapes (not shown), such as ovals, or other shapes could be used to capture shapes other than lines from the image 12.
The number of active optic segments in a sector, e.g., the four segments 502, 504, 506, 508 in sector 500, also has similar constraints. Smaller segments direct less light energy to the detector 80, but may provide more resolution of shape characteristics of the image 12′, whereas larger segments direct more light to the detector 80, thus are more easily detectable, but resolution decreases. For lower resolution applications or requirements, the sectors may not even need to be divided into segments, and this invention includes radial spatial light modulators in which each sector 500, 510, 520, 530, 540, 550, 560, 570, 580, 590, 600, 610, 620, 630, 640, 650 is not segmented, thus comprises a single active optic element for each sector. However, the same lower resolution effect can be achieved in the illustrated embodiment 50 in
In the preferred embodiment 50, each sector, e.g., sector 500, comprises four individually addressable, active optic segments, e.g., segments 502, 504, 506, 508, as shown in
To illustrate, suppose the image 12′ is a pattern of a plurality of parallel vertical lines intersecting a plurality of parallel horizontal lines to form a matrix of squares, as illustrated, for example, in
Also, light energy for the horizontal shape content of such large, small, or intermediate sized matrix square patterns would be incident on the respective inner, outer, or intermediate positioned segments of the horizontal sector 540. For example, in the image 12′ of
Further, any features of an image 12′ that have sizes over 50 percent of the size of image 12′, which light energy is incident on the center area portion 41, can either be captured and detected as an indicator of general brightness of the image 12′ for intensity control or calibration purposes or just ignored and not captured or detected at all, because there is little, if any, useable shape information or content in the light energy that comprises that 50 percent of the size of the image 12′. Likewise, the approximately 3⅛ percent of the size content of the image 12′ that is radially outward beyond the outer segments or sectors is not detected and can be ignored in this preferred configuration. The center 41 can be made optically active to capture light energy incident thereon, if it is desired to capture and detect such light energy for general brightness indication, intensity control, or calibration purposes, as will be understood and within the capabilities of persons skilled in the art, once they understand this invention. For example, if an image 12′ has a matrix of squares, which are so large that the distance between the vertical lines, which define the edges of the large squares, is over 50 percent of the width of the image 12′, there is little, if any, vertical shape content of practical use, and the light energy for that vertical shape content is incident on the center area portion 41. On the opposite end of the spectrum, if such an image 12′ has a matrix of squares, which are so small that the distance between the vertical lines, which define the edges of the small squares, is less than about 3⅛ percent of the width of the image 12′, there is also little, if any, vertical shape content of practical use, and the light energy for such vertical shape content is dispersed radially outward, beyond the outer segment 508 of sector 500. Of course, other configurations or scale segment sizes and combinations of the segmented radial SLM 50 could also be made and used within the scope of this invention.
The shape content detection according to this invention will be described in more detail below by use of the example automobile image 12′ of FIG. 5. However, it is helpful to understand at this point that, when the image 12′ is a matrix of squares, as described above, and when the light energy incident on the vertical sector 500 in the Fourier transform plane 32 is projected back into a spatial domain image, such spatial domain image will have been filtered to show only the vertical lines at the boundaries of the squares. No horizontal lines would appear in such spatial domain, filtered image, because the light energy with the horizontal shape content would have been substantially blocked or filtered out of the image. Further, if the squares of the matrix pattern are large, such as the squares 702 in
On the other hand, if the image 12′ has a matrix of very small squares 722, thus high spatial frequency, as shown in FIG. 7a and described above, then the light energy in the FT plane 32 is dispersed farther radially outward to be incident on the outer segment 508 and not on the inner segment 502. Therefore, the outer segment 508 of vertical sector 500 would have to be actuated to project such light energy of the vertical lines 724 of
In summary, for an image 12′ comprising a matrix of squares, as described above, actuation of the inner segment 502 of vertical sector 500 and getting vertical liens formed in the spatial domain, while actuation of the outer segment 508 as well as the intermediate segments 5604, 506, in the vertical sector 500 projects no vertical lines in the spatial domain, would show that the vertical shape content of the image has low spatial frequency characteristic of large squares 702 in FIG. 6a. Similar analysis with the horizontal sector 540 resulting in horizontal lines in the spatial domain from actuation of the inner segment 542, but not from actuation of the outer or intermediate segments 548, 546, 544, would show such horizontal lines 706 to also be low spatial frequency characteristic of large squares 702.
If analysis of other non-vertical and non-horizontal sectors 510, 520, 530, 550, 560, 580, 590, 600, 610, 620, 630, 640, 650 show no lines in the spatial domain from those angular orientations, then the recordable results confirm the shape content of the image 12′ to be only a smaller or larger spatial frequency at some or all of those angular orientations, then the recordable results would confirm some shape complexity in addition to the matrix of squares in the image 12′. Thus, shape information, including spatial frequency or scale (S), and intensity (I) at each angular orientation or rotation (R) can be acquired with the spatial light modulator 50 in the system of this invention.
In summary, for an image 12′ with the large square 702 matrix shown in
While the radial configuration of the active optic sectors with or without the multiple, active optic segments in each sector in the spatial light modulator 50 is a significant feature of this invention, persons skilled in the art of designing and fabricating spatial light modulators can readily understand how such a spatial light modulator 50 can be constructed and function, once they become familiar with the features and principles of this invention, and there are many known materials, fabrication techniques, and the like, known to persons skilled in the art that could be used to design, make, and use state-of-the-art spatial light modulators that are applicable to the specialized spatial light modulator embodiments of this invention. Therefore, a detailed recitation of such available materials is not necessary to enable a person skilled in the art to make and use this invention. Never-the-less, reference is now made to
As illustrated in
A transparent conductive layer 190 deposited on the front substrate 182 is connected by another lead 513 to another voltage V3. Therefore, a voltage can be applied across the portion of the liquid crystal material 180 that is sandwiched between the metal layer 186 and the transparent conductive layer 190 by, for example, making V1 positive and V3 negative and vice versa. Likewise, when a voltage can be applied across the portion of the liquid crystal material 180 that is sandwiched between the metal layer 188 and the transparent conductive layer 190 by, for example, making V2 positive and V3 negative and vice versa.
As mentioned above, the function of the respective segments 506, 508 is to rotate the plane of polarization of selective portions of the incident light beam 27(p) so that those portions of the light beam 21(p), which carry corresponding portions of the Fourier transform optic pattern 32, can be separated and isolated from the remainder of the light beam 27(p) for detection by the photodetector array 82 (FIG. 5). As understood by persons skilled in the art, there are a number of spatial light modulator variations, structures, and materials that can yield the desired functional results, some of which have advantages and/or disadvantages over others, such as switching speeds, light transmission efficiencies, costs, and the like, and many of which would be readily available and satisfactory for use in this invention. Therefore, for purposes of explanation, but not for limitation, the segmented radially spatial light modulator illustrated in
One example system, but certainly not the only one, can use a liquid crystal material 180 that transmits light 21(p) without affecting polarization when there is a sufficient voltage across the liquid crystal material 180 and to act as a 1/4-wave retarder when there is no voltage across the liquid crystal material. An untwisted crystal material 180 that is birefringent in its untwisted state can function in this manner. Thus, for example, when no voltage is applied across the liquid crystal material 180 in segment 508, there is no molecular rotation of the liquid crystal material 180 in outer segment 508, and the liquid crystal material in outer segment 108, with the proper thickness according to the liquid crystal manufacturer's specifications, will function as a ¼-wave plate to convert p-polarized light 27(p) incident on outer segment 508 to circular polarization as the light passes through the untwisted liquid crystal material 180. Upon reaching the metal layer 188, which is reflective, the light is reflected and passes back through the liquid crystal material to undergo another ¼-wave retardation to convert the circular polarization to linear polarization, but in the s-plane, which is orthogonal to the p-plane. The reflected light 61(s), therefore, has its plane of polarization effectively rotated by 90 degrees in relation to the incident light 27(p).
Meanwhile, if there is a sufficient voltage on, for example, the near outer segment 506, to rotate the long axes of the liquid crystal molecules into alignment with the direction of propagation of the incident light waves 27(p), thereby eliminating the birefringence of the liquid crystal material 180, then there is no change of the linear polarization of the light on either its first pass through the liquid crystal material 180 or on its second pass through the liquid crystal material after being reflected by metal layer 186. Consequently, under this condition with a voltage applied across the liquid material 180 in near outer segment 506, the reflected light 61(p) is still polarized in the p-plane, i.e., the same plane as the incident light 27(p).
Many liquid crystal materials require an average DC voltage bias of zero, which can be provided by driving the voltage V3 with a square wave function of alternating positive and negative voltages for equal times. Therefore, for no voltage across the liquid crystal material 180, the other voltages V1, V2, etc., can be driven in phase with equal voltages as V3. However, to apply a voltage across the liquid crystal material 180 adjacent a particular metal layer 186, 188, etc., to activate that particular segment 506, 508, etc., as described above, the respective voltage V1 or V2, etc., can be driven out of phase with V3. If the frequency of the square wave function is coordinated with the switching speed of the liquid crystal material 180, one-half cycle out of phase for a voltage V1, V2, etc., will be enough to activate the liquid crystal material 180 to rotate the plane of polarization of the light as described above.
As mentioned above, other alternate arrangements and known liquid crystal materials can reverse the results from an applied voltage. For example, a twisted liquid crystal material 180 may be used to rotate plane of polarization under a voltage and to not affect plane of polarization when there is no voltage.
Referring again primarily to
The reflected light 61(s) from the segmented radial SLM 50, e.g., light polarized in the s-plane reflected from an activated segment, as explained above, does not pass back through the polarizing beam splitter 70 along with p-polarized reflected light. Instead, the s-polarized reflected light 61(s) is reflected by the plane 72 in the polarizing beam splitter 70 to the detector 80 in the spatial domain. The lens 78 magnifies and focuses the isolated beam 61(s) in a desired size in the spatial domain on the detector array 82 of photodetector 80.
The photodetector array 82, as mentioned above, can be a 16×16 array of individual light sensors 84, such as charge coupled devices (CCDs), as shown in
The spatial filtering process described above and its characterization of the image 12 by shape content is illustrated in more detail in
This principle also facilitates design and fabrication of an effective segmented radial SLM 50, because, for every active optic sector, there can be an adjacent inactive sector or area available for placement of electrically conductive traces to the segments, as shown by reference back to
Returning now to
For example, shape content in the light energy characteristic of that incident on both the vertical 11.25° sector 500 centered at 0° as well as on the non-active area 581 centered at 180° can be detected by effectively activating the active optical segments 502, 504, 506, 508 of sector 500. Shape content in the light energy characteristic of that incident on the 11.25° sector 590 centered at 191.25° as well as on the non-active area 501 centered at 11.25° can be detected effectively by activating the active optic segments of sector 590, because the active optic sector 590 is centered diametrically opposite the non-active area of 11.25°. Shape content in the light energy characteristic of that incident on either the 11.25° sector 510 centered at 22.5° or the non-active area 591 centered at 202.5° can be detected by activating the active optic segments of sector 510. Shape content in the light energy characteristic of that incident on either the 11.25° non-active area centered at 33.75° or active sector 600 centered at 213.75° can be detected by activating the active optic segments of sector 600, which is centered diametrically opposite 33.75° at 213.75°. Shape content in the light energy characteristic of that incident on either the 11.25° sector 520 centered at 45° or non-active area 601 centered at 225° can be detected by activating the active optic segments of sector 520. Shape content in the light energy characteristic of that incident on either the 11.25° non-active area 521 centered at 56.25° or the active sector 610 centered at 236.25° can be detected by activating the active optic segments of sector 610, which is centered diametrically opposite 56.25° at 256.25°. Shape content in the light energy characteristic of that incident on either the 11.25° sector 530 centered at 67.5° or the non-active area 611 centered at 247.5° can be detected by activating the active optic segments of sector 530. Shape content in the light energy characteristic of that incident on either the 11.25° non-active area 531 centered at 78.75° or active sector 620 centered at 258.75° can be detected by activating the active optic segments of sector 620, which is centered diametrically opposite 78.75° at 258.75°. Shape content in the light energy characteristic of that incident on either the 11.25° sector 540 centered at 90° or non-active area 621 centered at 270° can be detected by activating the active optic segments of sector 540. Shape content in the light energy characteristic of that incident on either the 11.25° non-active area 541 centered at 101.25° or the active sector 630 centered at 281.25° can be detected by activating the active optic segments of sector 630, which is centered diametrically opposite 101.25° at 281.25°. Shape content in the light energy characteristic of that incident on either the 11.25° sector 550 centered at 112.5° the diametrically opposite portion of non-active area 631 that is centered at 292.5° can be detected by activating the active optic segments of sector 550. Shape content in the light energy characteristic of that incident on the 11.25° sector 560 centered at 123.75°. The diametrically opposite portion of non-active area 631 that is centered at 303.75° can be detected by activating the active optic segments of sector 560. Shape content in the light energy characteristic of that incident on the 11.25° non-active area 561 centered at 135° or active sector 640 centered at 315° can be detected by activating the active optic segments of sector 640, which is centered diametrically opposite 135° at 315°. Shape content in the light energy characteristic of that incident on the 11.25° sector 570 centered at 146.25° or non-active area 641 centered at 326.25° can be detected by activating the active optic segments of sector 570. Shape content in the light energy characteristic of that incident on the 11.25° non-active area 571 centered at 157.5° or active sector 650 centered at 337.5° can be detected by activating the active optic segments of sector 650, which is centered diametrically opposite 157.5° at 337.5°. Finally, shape content in the light energy characteristic of that incident on the 11.25° sectors 580 centered at 168.75° or non-active area 651 centered at 348.75° can be detected by activating the active optic segments of sector 580.
While it would be unnecessarily cumbersome to illustrate and describe the shape detecting and characterizing functionality of all the active optic segments of all the sectors 500, 510, 520, 530, 540, 550, 5.60, 570, 580, 590, 600, 610, 620, 630, 640, 650, it may be helpful for an understanding of the invention to illustrate and describe the functionality and results of activating several representative examples of the active optic segments in the active optic area 54. Therefore,
The light energy bands 34, when reflected by the activated outer segment 508, are filtered through the polarizing beam splitter 70 and projected in the filtered optic pattern 60, which is comprised primary of vertical lines or bands 62 of light energy illustrated diagrammatically in
To illustrate further, the near inner segment 504 of active optic sector 500 is shown in
Another example angular orientation of light energy 34 from the FT optic pattern 32 is illustrated by
Capture and detection of horizontal portions of lines, edges, and features 68, 68′ of the image 12′ of respective spatial frequencies, if present in the image 12′, is accomplished by activation of one or more segments 542, 544, 546, 548 of horizontal sector 540, which is oriented 90° from the vertical 0°. The portion of the light energy 34 that is reflected by each activated segments 542, 544, 546, 548 of the horizontal sector 540 is characteristic of all of the substantially horizontal features, parts, and lines 68 of the respective spatial frequencies in the image 12′ that correspond to the light energy, if any, that is incident on those segments in the FT plane 32, as shown in FIG. 13b. Some curved features, parts, or lines in the image 12′ have portions or line segments 68′ that are also substantially horizontal, so those horizontal portions or line segments 68′ also contribute to the light energy 34 that gets reflected by the horizontal sector 540 in FIG. 13a. The bands 62 of light energy in the filtered pattern 60, shown in
One more example activated segment 598 in sector 590, is illustrated in
Referring again to
It should be clear by now that any particular angular orientation R of segments of sectors in the active optic area 54 will allow detection of all the shape characteristics of image 12′ that have substantially that same angular orientation R. It should also be clear that radial outward spacing or scale (S) of the segments relates to spatial frequency of such shape characteristics. Thus, all of the shape characteristics of the image 12′ can be detected by detecting the bands 62 of the respective filtered patterns 60 with the segments at all angular orientations. However, as mentioned above, it is sufficient for most purposes to detect some, preferably most, but not necessarily all, of the shape characteristics of the image 12′ by choosing to detect the light energy bands 34 of filtered patterns 60 at certain selected increments of or angular orientation or rotation R. Obviously, the bigger the increments of angular orientation of the sectors where light energy bands 34 are detected, the less precise the detected shape characteristics or contents of the image 12′ will be. On the other hand, the smaller the increments of angular orientation, the more data that will have to be processed. Therefore, when selecting the angular increments of sectors at which light energy bands 34 will be detected and recorded, it may be desirable to strike some balance between preciseness of shape characteristics needed or wanted and the speed and efficiency of data processing and storage required to handle such preciseness. For example, but not for limitation, it is believed that detection and recording of the shape characteristics at angular increments of in a range of about 5 to 20 degrees, preferably about 11.25-degrees, will be adequate for most purposes. Also, the angular area of detection can be varied. For example, even if active optic sectors are oriented to detect shape characteristics at angular increments of 11.25°, the active optic areas could be narrow, such as in a range of 3° to 8°, more or less, which would filter out some of the optic energy from the FT optic pattern 32 between the sectors. However, such loss of light energy from non-active areas between sectors or other radially extending sensors, as described elsewhere in this specification, may not be detrimental to shape characterization by this invention, depending on specific applications of the technology to particular problems or goals.
Instead of the radially extending, wedge-shaped active optic sectors and segments of sectors described above, an alternate configuration can be comprised of radially extending, rectangular-shaped active optic modulators as illustrated diagrammatically in FIG. 15. These rectangular-shaped modulators 500′, 510′, 520′, 530′, 540′, 550′, 560′, 570′, 580′, 590′, 600′, 610′, 620′, 630′, 640′, 650′ can be at the same or different angular orientations as the wedge-shaped sectors described above, and each angular orientation can comprise several rectangular, active optic segments, such as segments 502′, 504′, 506′, 508′ of the modulator 500′. This arrangement does not capture as much of the light energy of an incident FT optic pattern 32 (
Another, albeit less efficient embodiment, is illustrated in
While the reflective spatial light modulator structure described above in connection with the cross-sectional view of
In the description above of the preferred implementation of this invention, the shape content of a desired angular orientation (R) and scale (S) of an image is captured by masking or blocking all other light in the FT plane 32 from reaching the detector 80 so that only the light energy from that angular orientation (R) and scale segment of the FT plane 32 gets projected back into the spatial domain for detection. However, persons skilled in the art will recognize that this invention can also be practiced in the negative. In other words, instead of actuating the one or several segments and/or sectors to get shape content relevant to the angular orientation or rotation (R) and/or radial distance (S) of particular sectors and/or segments, as described above, it would also be feasible to actuate all of the other sectors and segments in the active optic area 54 and not actuate the specific sector and/or segments in order to get a negative or inverse of the shape content of the image. This procedure can be repeated for all of the desired angular (R) and/or scalar (S) sectors and segments so that the composite of information regarding light energy distribution collected and recorded represents a negative or inverse of all of the shape content of an image 12′.
For example, referring back to FIGS, 6a-c, the negative or inverse of the vertical shape content of
To explain further, a negative of the spatially filtered image 60 of the automobile illustrated in
Again, as with the positive spatial images of the shape content, such negative or inverse spatial images can be detected at 80 (
The accuracy, versatility, and efficiency of shape characterizing, processing, storing, searching, comparing, and matching images according to this invention can be enhanced by some pre-processing of the images 12, 14, . . . , n when creating the optical patterns for the images 12′, 14′, . . . , n′ at the SLM 26 in FIG. 5. One particularly beneficial method of such pre-processing is “ghosting” the image to allow more light energy into the optic pattern 12′, thus also allowing more light energy into the FT optic pattern 32.
With reference now to
After the image 800 is converted to an optic pattern of the edge content 802 of the image 800, as illustrated in
The wider radial dispersion of light energy in FT optic pattern 32 due to the higher spatial frequency content in the ghosted image 802′ of
The ghosting process is quite simple and can be scaled to achieve a desired result. Essentially, a software program can simply be applied to reproduce each pixel of an image at selected locations, at selected distances, and at selected angular orientations in relation to such pixel, as illustrated in the simple example of the dot 800 in
The ghosting process of this invention can also be applied to images for which edges have not been found or produced, as described above. However, more pixel processing by the computer 20 or other processor would be required, and resulting shape resolution may not be as sharp.
Since these and numerous other modifications and combinations of the above-described method and embodiments will readily occur to those skilled in the art, it is not desired to limit the invention to the exact construction and process shown and described above. For example, Accordingly, resort may be made to all suitable modifications and equivalents that fall within the scope of the invention as defined by the claims which follow. The words “comprise,” “comprises,” “comprising,” “include,” “including,” and “includes” when used in this specification and in the following claims are intended to specify the presence of stated features or steps, but they do not preclude the presence or addition of one or more other features, steps, or groups thereof.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US3689772||Aug 18, 1971||Sep 5, 1972||Litton Systems Inc||Photodetector light pattern detector|
|US3771124 *||Jan 21, 1972||Nov 6, 1973||Sperry Rand Corp||Coherent optical processor fingerprint identification apparatus|
|US4360799||May 22, 1980||Nov 23, 1982||Leighty Robert D||Hybrid optical-digital pattern recognition apparatus and method|
|US4462046 *||Jul 2, 1982||Jul 24, 1984||Amaf Industries Incorporated||Machine vision system utilizing programmable optical parallel processing|
|US4735487||Nov 12, 1985||Apr 5, 1988||Jenoptik Jena Gmbh||Optical apparatus for analyzing two-dimensional input objects|
|US4923297 *||Jun 23, 1986||May 8, 1990||Eyedentify, Inc.||Optical alignment system|
|US5097352||Sep 27, 1989||Mar 17, 1992||Olympus Optical Co., Ltd.||Optical system including liquid crystal devices|
|US5177559 *||May 17, 1991||Jan 5, 1993||International Business Machines Corporation||Dark field imaging defect inspection system for repetitive pattern integrated circuits|
|US5216541 *||May 8, 1990||Jun 1, 1993||Sumitomo Cement Company Ltd.||Optical associative identifier with real time joint transform correlator|
|US5264912 *||Feb 7, 1992||Nov 23, 1993||Tencor Instruments||Speckle reduction track filter apparatus for optical inspection of patterned substrates|
|US5659637 *||Jun 20, 1996||Aug 19, 1997||Optical Corporation Of America||Vander lugt optical correlator on a printed circuit board|
|US6424449||Apr 19, 2000||Jul 23, 2002||Olympus Optical Co., Ltd.||Optical information processing apparatus for image processing using a reflection type spatial light modulator|
|JP2001142029A||Title not available|
|JPH0291607A||Title not available|
|WO2000017809A1||Sep 24, 1998||Mar 30, 2000||Meirion Francis Lewis||Improvements relating to pattern recognition|
|1||Anonymous; "Holographic Image Converter for Optical Information Processing", IBM Technical Disclosure Bulletin, vol. 27, No. 7A, Dec. 1984, pp. 3745-3746.|
|2||D. Casasent et al., "Diffraction Pattern Sampling Using a Computer-Generated Hologram"; Applied Optics, vol. 25, No. 6, Mar. 1986; pp. 983-989.|
|3||Z. Chen et al., "Hybrid Optical/Digital Access Control Using Fingerprint Identification"; Optical Engineering, vol. 34, No. 3, Mar. 1995; pp. 834-838.|
|U.S. Classification||382/211, 382/280, 708/403|
|International Classification||G06F15/00, G06K9/76, G06T7/00, G06K9/58, H04N5/92, H04N5/76, G06K9/52, G06F17/30, G06K9/74|
|Cooperative Classification||G06K9/748, G06K9/58, G06F17/30247|
|European Classification||G06K9/58, G06K9/74S, G06F17/30M1|
|Apr 18, 2014||REMI||Maintenance fee reminder mailed|
|Sep 3, 2014||FPAY||Fee payment|
Year of fee payment: 8
|Sep 3, 2014||SULP||Surcharge for late payment|
Year of fee payment: 7