WO1999060776A1

WO1999060776A1 - Grayscale enhancement system and method

Info

Publication number: WO1999060776A1
Application number: PCT/US1999/011344
Authority: WO
Inventors: Daniel L. Allen
Original assignee: Oak Technology, Inc.
Priority date: 1998-05-21
Filing date: 1999-05-21
Publication date: 1999-11-25
Also published as: US6266154B1

Abstract

A grayscale enhancement system which operates on color or monochrome source halftone images to produce output halftone cells of the same visual weight but spatially distributed to reduce artifacts caused by limited grayscale levels. By operating upon a set of input halftone cells surrounding each output pixel cell with a set of logic operations implementing unique algorithms, the system re-distributes the weight of the halftone cells characterizing inferred-halftone cells. A rendering subsystem, responsive to the inferred-halftones, then produces signal for driving an output device to the best of the output device's ability.

Description

GRAYSCALE ENHANCEMENT SYSTEM AND METHOD

FIELD OF INVENTION This invention relates to a grayscale enhancement method and system including a unique routine which eliminates or reduces the mottling effect of present halftoning techniques.

BACKGROUND OF INVENTION Shades of gray on computer printers are achieved by a process called halftoning. In the halftoning process, halftone cells are created each including a number of pixels. By making all the pixels white, a white spot is created and, conversely, by making all the pixels black, a black spot is created. By making only a small percentage of the pixels black in a cell, the spot, to the eye, appears light gray and conversely by making a large percentage of the pixels black in a cell, the spot appears dark gray. In general, the number of gray levels available is a function of the number of pixels in a halftone cell. However, a given printer's capacity is a function of the number of pixels per inch and the size of the halftone cell. For example, if a cell includes 9 pixels, there are 10 gray levels possible while if the cell includes 256 pixels, there are 257 possible gray levels. Thus, the finer the halftone screen, the fewer number of pixels in each cell and therefore the fewer gray levels available. So, a 600 dpi printer at a screen frequency of 200 lines per inch (lpi) has 10 possible gray levels while the same printer at a screen frequency of 120 lpi has 26 possible gray levels. Therefore, to achieve more gray levels, screen frequency

and hence image resolution suffers.

So, the goal has been to increase the printer's image resolution on 600 dpi printers and achieve more gray levels at the same screen frequency. That is, a 600 dpi printer at 141 lpi results in 19 gray levels but with super dot techniques can produce 145 gray levels at 141 lpi.

Even with such advancements in technology and the use of screen angling and spot spacing variation techniques, however, a phenomenon known as mottling occurs which makes photographic and other images appear to have bad complexions. See, in general, Real World

Scanning and Halftones, Blatner and Roth (1993, Peachpit Press).

Current processes for correcting the mottling effect include techniques that unfortunately add significant storage requirements to the printer system.

SUMMARY OF INVENTION

It is therefore an object of this invention to provide a grayscale enhancement method and system which eliminates or reduces the mottling effect of present halftone techniques.

It is a further object of this invention to provide such a grayscale enhancement method and system which corrects the mottling artifact without adding to system storage requirements.

It is a further object of this invention to provide such a grayscale enhancement method and system which results in magazine quality printing using standard laser printers.

It is a further object of this invention to provide such a grayscale enhancement method and system which can be implemented in existing printer software and integrated with other text enhancement techniques to produce a total system for producing quality text, line art, and photo images.

It is a further object of this invention to provide such a grayscale enhancement method and system which does not require significant software modifications or increases in processing requirements for the printer controller.

It is a further object of this invention to provide such a grayscale enhancement method and system which does not add any significant cost to the printer controller.

This invention results from the realization that the mottling effect of current halftoning techniques can be corrected, not by techniques which add significant storage requirements to a printer's controller, but by replacing full size, widely spaced black pixels with closer spaced black pixels of a smaller size such that the total percentage of darkness is maintained but the spacing between the adjacent black pixels is reduced thereby eliminating or reducing the mottling effect of current halftoning techniques.

This invention features a grayscale enhancement method comprising the steps of analyzing the state of selected pixels surrounding a subject pixel in a bit map; determining the number of selected pixels including the subject pixel that are black; and printing, at a location corresponding to the location of the subject pixel in the bit map, a fractional pixel related in size to at least the computed number of pixels which are black.

The step of analyzing preferably includes generating a window which moves about the bit map. The window is typically nine by ten pixels in area. The selected surrounding pixels are usually located at a predetermined set of vectors with respect to the subject pixel within the window. The predetermined set of vectors then each have a length n, for example 3 /2 pixels and each vector preferably extends to a different corner of the window.

A number, e.g. 4=m pixels are selected as the surrounding pixels and printing includes driving a printer modulator to print a pixel of a size approximately x/(m+ l) where x is the number of black pixels present in the set including the m surrounding pixels and the subject pixel.

The grayscale enhancement method of this invention comprises analyzing the state of a plurality of pixels in a portion of a bit map; generating a number of fractional size pixels for each black pixel of the plurality; and printing the fractional size pixels. The plurality typically contains m pixels and each fractional size pixel has a size 1/m. The number of fractional size pixels is then m.

The grayscale enhancement system of this invention includes means for analyzing the state of selected pixels surrounding a subject pixel in a bit map; means for determining, from the selected pixels and the subject pixel, the number of black pixels; and means for printing, at a location corresponding to the location of the subject pixel in the bit map, a fractional pixel related in size to at least the computed number of pixels which are black. The means for analyzing includes means for generating a window which moves about the bit map. The window is preferably nine by ten pixels in area. The means for analyzing further includes means for selecting surrounding pixels that are located at a predetermined set of vectors with respect to the subject pixel within the window. The predetermined set of vectors each have a length n, e.g.

3Λ/2 and each vector extends to a different corner of the window. Printing includes driving a printer modulator to print a pixel of a size approximately x/(m+ l) where x is the number of black pixels present in the set including m surrounding pixels and the subject pixel m may equal

4.

The grayscale enhancement system of this invention includes means for analyzing the state of a plurality of pixels in a portion of a bit map; means for generating a number of fractional size pixels for each black pixel of the plurality; and means for printing the fractional size pixels. The plurality contains m pixels and each fractional size pixel has a size 1/m. The number of fractional size pixels is m.

DISCLOSURE OF PREFERRED EMBODIMENT

Other objects, features and advantages will occur to those skilled in the art from the following description of a preferred embodiment and the accompanying drawings, in which:

Fig. 1 is an illustration of prior art halftone cells used in conjunction with prior art halftoning techniques;

Fig. 2 is an illustration of prior art "super dot" halftone cells;

Fig. 3 is conceptual illustration of the result of using the grayscale enhancement method and system of this invention for lightest shade of gray;

Fig. 4 is an illustration of a prior art darker shade of gray in a "super dot" halftone cell;

Fig. 5 is a conceptual illustration of the result of using the grayscale enhancement method and system of this invention for the shade of gray represented by the "super dot" halftone cell of Fig. 4;

Fig. 6 is an illustration of still a darker shade of gray for a prior art "super dot" halftone cell;

Fig. 7 is a conceptual illustration of the result of using the grayscale enhancement method and system of this invention for the prior art "super dot" halftone cell shown in Fig. 6;

Fig. 8 is an illustration of a still darker shade of gray for a prior art "super dot" halftone cell;

Fig. 9-20 are illustrations of the moving window and analysis computation steps of the grayscale enhancement method of this invention;

Fig. 21 is a block diagram showing the primary components of the grayscale enhancement system of this invention;

Fig. 22 is a flow chart of the grayscale enhancement routine operable in the system shown in Fig. 21 in accordance with this invention; and

Figs. 23-24 are conceptual illustrations of the windows created in accordance with the subject invention illustrating halftone images and text character or line edges, respectively.

Halftone cell 10, Fig. 1 has one black pixel 2 out of 18 total pixels (a typical halftone dot for a 600 dpi laser printer with a frequency of 141 lpi) and adjacent halftone cell 14 also has single black pixel 6 out of 18 total pixels in accordance with the prior art. By adding additional black pixels in each cell, different shades of gray are achieved. In this example, there are 19 possible shades of gray as shown in Fig. 1.

Additional gray shade levels are achieved in accordance with the prior art "super dot" methodology by combining four cells, Fig. 2. Groups of halftone dots are treated as a "super- dot" so that four dots with 18 bits each becomes a "super dot" with 72 bits and, theoretically, 73 shades of gray.

At the lighter shades of gray, however, the distance d₂ between black pixels 12 and 16 is twice the distance d[ shown in Fig. 1 between black pixels 2 and 6. This large distance results in the mottling effect discussed in the Background of the Invention above. In summary, the "super dot" methodology, Fig. 2, adds additional shades of gray, but suffers from the mottling phenomenon at the lighter shades of gray.

A quite different approach to distribute the grays is shown in U.S. Patent No. 5,533,150 in which an isolated black pixel is distributed into smaller sub-pixels, with a similar total weight, however the distribution is done within the same halftone cell. In this approach, only isolated pixels are treated, whereas the objective of the present invention is to not only modify the distribution of isolated black pixels but also distribute the weight of halftone cells comprised of several pixels of black. Further, the subject invention distributes the weight of these halftone dots over several halftone cells rather than within each cell therefore maintaining the true halftone relationship and therefore, increasing the number of gray levels. In contrast, the technique shown in U.S. Patent No. 5,533,150 does not increase the number of gray levels.

As a conceptual matter, in this invention, however, full pixels 12 and 16, Fig. 2 are replaced with smaller, closer spaced pixels 30, 32, 34, and 36; and 38, 40, 42, and 44, respectively, Fig. 3. Note that the distance d₃ between fractional size pixels 30 and 32 is significantly less than the distance d₂, Fig. 2, between full size pixels 12 and 16 and also that the gray level appears the same to the eye since fractional pixels 30-36, Fig. 3, are each one fourth the size of pixel 12, Fig. 2.

For the next darker shade of gray, Fig. 4, full size black pixels 46 and 48, Fig. 4 in halftone cell 49 are replace with fractional half size pixels 50, 52, 54, and 56, Fig. 5. For the next darker shade of gray, Fig. 6, full size black pixels 60, 62, and 64 are replaced with fractional three quarter size pixels 70, 72, 74, and 76, Fig. 7.

The grayscale shade shown in Fig. 8 is not altered since the spacing between pixels 80, 82, 84, and 86 is already optimized. Another way to consider this case is that full size pixels 80- 84 are "replaced" with pixels of four quarters the size.

The process then continues in a like fashion for the darker grayscale shades.

Although within the scope of this invention and although such a template matching process could continue to replace full size pixels representing every possible shade of gray in a bit map with the appropriate fractional sized pixels, it is much more convenient and preferred to use a bit map window based process in accordance with this invention as discussed with reference to Figs. 9-20.

Accordingly there are some means for analyzing the state of selected pixels surrounding a subject pixel in a bit map such as window subsystem 170, Fig. 9 which forms a moving window within the source pixel-map. The window is typically an even number of pixels in the horizontal direction of the output scan and an odd number of pixels in the vertical direction. In the preferred embodiment, the window is 9 by 10 pixels with a total of 90 input pixels. The process of forming the moving window requires line stores if the source data is presented in raster-scan fashion, as is typical. In the case of a 9-line window, eight prior lines are stored. The ninth line is the currently inputting line and need not be stored. These line stores, known in the art, are a major portion of the gate count and system cost and can be lowered if multiple scan lines (not necessarily all 9) are accessed simultaneously from a source RAM. The exact size and shape of the window can be modified to accommodate the size and angle of the source halftone line screen and the systems ability to enhance several line screens. Further, it is desired to use a 45-degree screen rather than a 90-degree screen as discussed in Figs. 1-8. The screen angle is important due to the fact that the human eye can see horizontal and vertical lines particularly well. For this reason, a pattern of halftone dots arranged at an angle of 90 degrees (vertical and horizontal rows of halftone dots) is most easily recognized and therefore tends to be visible. The least recognizable line screen pattern for printing is that with a raster angle of 45 degrees. In the preferred embodiment, shown in Fig. 9, the typical pixel arrangement for a 45 degree screen with the window subsystem is shown by selected pixels 174, 172, 176, 178, and 180. This set of selected pixels are intentionally off-set to the left in window 170 for reasons to be explained later.

Window 170, Fig. 9, traverses a bit map to analyze the state of selected pixels 172, 174, 176, and 178 surrounding "subject" pixel 210. Each pixel in the bit map eventually becomes a "subject" pixel due to movement of window 170 about the bit map. In this case, subject pixel 210 is white and selected surrounding pixels 172, 174, and 176 178 are white. Pixel 180, however, is black. In the preferred embodiment, window 170 is 9 by 10 pixels in area and the selected surrounding pixels 172, 174, 176, and 180 are all located at a predetermined set of vectors 182, 184, 186, and 188, respectively, with respect to subject pixel 210. Each vector has a length of 3^/2 pixels and is at different angle from the subject pixel within the window (e.g.,

45°, 135°, 225°, and 315°).

The number of black pixels present in the set comprising the subject pixel and the selected surrounding pixels is then computed. In this case, only one pixel is black, namely pixel 180. Then, instead of printing full size pixel 180, the modulator of the printer is driven to print fractional size pixel 200, Fig. 10 of approximately one fifth or twenty percent the size of full pixel 180. Fractional pixel 200 is printed on the printed page at a location corresponding to where original full size subject pixel 180, Fig. 9, was to be printed. The software, however, places fractional size pixel 200 in the center of window 170 as shown in Fig. 10.

In Fig. 11, window 170 has traversed the bit map and is now analyzing new subject pixel 211. Prior black pixel 180 is now one of the selected pixels surrounding new subject pixel 211 since it is located at vector 188 with respect to new subject pixel 211.

Again, there is now one black pixel out of five and fractional pixel 212, Fig. 12, twenty percent of the size of full pixel 180, is printed at a location corresponding to new subject pixel 210. Window 170 continues to move until, as shown in Fig. 13, black pixel 180 is now the center "subject" pixel in window 170. Since there is again one black pixel out of five in window

170, fractional pixel 216, Fig. 14 is printed.

This process of analyzing the state of selected pixels surrounding a subject pixel within the window, computing the number of black pixels present for the total of the subject pixel combined with a selected surrounding pixels, and sending a signal to the modulator of a printer to print a fractional pixel continues until the window has completely traversed the bit map and each pixel and its surrounding selected pixels have been analyzing as shown in Figs. 15-20. In Fig. 15, the window 170 has traversed the bit map and is analyzing subject pixel 215 when dark pixel 180 is at vector 184 from subject pixel 215. As a result, fractional twenty percent pixel 220, Fig. 16 is printed. In Fig. 17, window 117 has moved across the bit map and is analyzing subject pixel 218 and black pixel 180 now lies at vector 182 with respect to subject pixel 218. As a result, fractional twenty percent pixel 222 is printed, Fig. 18.

The result is that full size black pixel 180, Fig. 19, is "replaced" with one fifth size pixels 200, 212, 216, 220, 222, Fig. 20. The result is the same grayscale level since, when combined, the five fractional pixels total the size of full size pixel 180 and yet the amount of white space between these new fractional size pixels is greatly reduced thereby eliminating the mottling effect noticeable to the eye when looking at a printed page resulting from the lighter shades of gray achievable using prior art halftoning techniques.

The primary components of this invention as shown in Fig. 21 include a window subsystem 254, which operates on the controller's output pixel bit map 258 and stores multiple scan lines of the serial input pixel bit-map. The window subsystem feeds enhancement selector

253 that determines whether the pixels in the window subsystem are likely to be a halftone or text/line-art. If enhancement selector 253 determines that the contents of the window subsystem

254 is likely to be text or line-art, then text and line-art enhancement subsystem 251 can be optionally used to operate on the window subsystem contents. When enhancement selector 253 determines that contents of window subsystem 254 is likely to be a halftone, then the grayscale enhancement subsystem 252 operates on the window subsystem 254. Operating grayscale enhancement subsystem 252 is responsive to enhancement selector 253 and window subsystem

254. The output of grayscale enhancement subsystem 252 is a signal presented via look-up table

225 to pixel generator 256 such as a modulator to print the appropriate fractional size pixels at their proper location. Generator 256 drives a printing device 250 such as a laser or ink jet print engine capable of sub-pixel modulation. Generator 256 may be a sub-pixel modulator such as

Xerographic Laser Image's modulator model no. XLI-2050 or Analog Devices' AD9561 Pulse

Width Modulator. The output of controller 258 is a standard serial video data stream which a page description language has generated. Grayscale enhancement subsystem 252 of this invention operates on that video data stream and is processed on-the-fly while the page is being printed. The output of subsystem 252 is an enhanced digital video signal presented to modulator

256 which drives the laser in the print engine. The means for determining from the selected pixels and the subject pixel, the number of black pixels may be grayscale enhancement system

252, Fig. 21 which operates in accordance with the flow chart shown in Fig. 22. As discussed above, a window is made to traverse a bit map to analyze each pixel which at one point in time is a "subject" pixel and certain selected pixels surrounding the subject pixel, step 260, to determine their state. In step 262, the number of pixels which are black of the subject pixel and the selected surrounding pixels is computed, and then a signal is generated, step 264, to drive the printer to print an appropriate fractional pixel related in size to the computed number of pixels which are black at the location of the subject pixel in the bit map. The window is now moved to the next pixel in the bit map which becomes a "subject" pixel, step 266 and processing returns to step 260.

Note, however, that the grayscale enhancement methodology described above is not always implemented for every pixel in a bit map. For example, text is not to be "enhanced" in accordance with this invention and therefore a routine, described below, is actually a precursor to step 260, Fig. 22.

It is not desirable to modify pixels using the halftone enhance algorithm of this invention when text or line-art is to be printed. Halftone dots are always separated by a distance "d" from each other and that distance forms the basis for making a decision by the examination of the window for multiple pixels within the halftone cell that are separated by a distance less than "d" . When such an occurrence is present, the halftone enhancement of this invention is turned off, and if present in the system, an edge enhancement algorithm is turned on. Figure 23 shows typical areas of the window where black pixels may be present that give a high likelihood of halftone images being present. Figure 24 shows the content of a typical window when a text character or line edge is present.

The undesirable "mottling" effect in images occurs mainly in a highlight area (light areas of images such as facial tones) of photo images. When highlight areas of images are detected, the halftone enhancement algorithm of this invention is turned on. This condition is readily determined by checking certain black pixels within a halftone cell and determining if the state has exceeded a predetermined threshold.

The following listing describes the actual logical operations of the "enhancement selector" in the preferred embodiment for the "left window" analysis to determine whether halftone dots are present. The equivalent analysis is done for the "right window" analysis. If the logical operations detect a halftone in either the "left window" analysis. If the logical operations detect a halftone in either the "left window" OR "right window" , then halftone dots are assumed to be present thus allowing the rendering system to execute its sub-pixel distribution for "grayscale enhancement" to "smooth" the image by increasing the number of gray levels in an image and therefore significantly reducing the "mottling" effect in photo images. The 1-bit grayscale enhancement entity

0 1 2 3 4 5

0 000, 220, 221, 222, 223, 224,

1 220, 225, 226, 227, 228, 229,

2 221, 226, 230, 231, 232, 233,

3 222, 227, 231, 234, 235, 236,

4 223, 228, 232, 235, 237, 238,

5 224, 229, 233, 236, 238, 000,

LIBRARY leee;

USE leee. std_logιc_1164.ALL;

ENTITY graylb t IS

PORT (bmcO IN STD_LOGIC_VECTOR(0 TO -- Bitmap columnO — RIGHT

MOST bmcl IN STD_LOGIC_VECTOR(0 TO Bitmap columnl bmc2 IN STD_LOGIC_VECTOR(0 TO Bitmap column2 bmc3 IN STD_LOGIC_VECTOR(0 TO Bitmap column3 bmc4 IN STD_LOGIC_VECTOR(0 TO Bitmap column4 bmc5 IN STD_LOGIC_VECTOR(0 TO Bitmap column5 bmc6 IN STD_LOGIC_VECTOR(0 TO Bitmap columnβ bmc7 IN STD_LOGIC_VECTOR(0 TO Bitmap column7 bmc8 IN STD_LOGIC_VECTOR(0 TO Bitmap columnδ bmc9 IN STD LOGIC VECTOR (0 TO Bitmap column9 — LEFT

MOST elk IN STD_LOGIC; — Clock

Gcase OUT STD_LOGIC_VECTOR(0 TO 18);

GrayDetected : OUT STD_LOGIC; Set when we should over-ride the edge detection with a grayscale case.

Gsymmetry: OUT STD_LOGIC

end graylbit; ARCHITECTURE structural OF graylbit IS

SIGNAL LeftCount : STD_LOGIC_VECTOR ( 0 TO 5); -- value is not binary encoded.

SIGNAL RightCount : STD_LOGIC_VECTOR (0 TO 5); — same as above

SIGNAL LeftGrayDetected : STD_LOGIC;

SIGNAL RightGrayDetected : STD_LOGIC;

SIGNAL iGrayDetected : STD_LOGIC; — need an internal version of this signal

SIGNAL iGsymmetry : STD_LOGIC;

SIGNAL xGsymmetry : STD_L0GIC;

SIGNAL iGcase : STD LOGIC VECTOR(0 to 18);

SIGNAL LA2 STD_LOGIC SIGNAL LAI STD_LOGIC SIGNAL LAO STD_LOGIC SIGNAL LB2 STD__LOGIC SIGNAL LB1 STD_LOGIC SIGNAL LBO STD_LOGIC SIGNAL LC4 STD_LOGIC SIGNAL LC3 STD_LOGIC SIGNAL LC2 STD_LOGIC SIGNAL LCI STD__LOGIC SIGNAL LCO STD LOGIC

SIGNAL RA2 STD_LOGIC SIGNAL RA1 STD_LOGIC SIGNAL RAO STD_LOGIC SIGNAL RB2 STDJLOGIC SIGNAL RBI STD_LOGIC SIGNAL RBO STD_LOGIC SIGNAL RC4 STD_LOGIC SIGNAL RC3 STD_LOGIC SIGNAL RC2 STD LOGIC SIGNAL RC1 : STD_LOGIC; SIGNAL RCO : STD_LOGIC;

BEGIN — {

-- bmc 9 8 7 6 5 4 3 2 1 0

0 o o o

1 1 r o o o l r

2 o o o

— 3 o o o o o o

4 o o o L R o o o

-- 5 o o o o o o

-- 6 o o o

7 1 r o o o l r

-- 8 o o o

LeftGrayDetected <= '0' WHEN (

-- If certain pixels are black, then no grayscale enhancement. bmc6(0) = '1' OR bmc5(0) = ' 1' OR bmc4(0) = '1' OR bmc6(l) = '1' OR bmc5(l) = '1' OR bmc4(l) = ' 1' OR bmc6(2) = '1' OR bmc5(2) = '1' OR bmc4(2) = ' 1' OR bmc6(6) = '1' OR bmc5(6) = '1' OR bmc4(6) = '1' OR bmc6(7) = '1' OR bmc5(7) = '1' OR bmc4(7) = '1' OR bmc6(8) = '1' OR bmc5(8) = ' 1' OR bmc4(8) = '1' OR bmc9(3) = ' 1' OR bmc8(3) = '1' OR bmc7(3) '1' OR bmc3 (3) = '!' OR bmc2(3) = '1' OR bmcl(3) = '1' OR bmc9(4) = '1' OR bmc8(4) = '1' OR bmc7(4) = ' 1' OR bmc3(4) OR bmc2(4) = '!' OR bmcl(4) = '!' OR bmc9(5) = '1' OR bmc8(5) = '1' OR bmc7(5 ' 1' OR bmc3(5) = '1' OR bmc2(5) = '!' OR bmcl(5) = '1' OR

(bmc5 (4) 0' AND -- if middle pixel is white...

-- If just one corner is set, make sure that no adjacent pixels are set. If so this must be text! (bmc8(l) = '1' AND (bmc9(0) = '1' OR bmc8(0) = '!' OR bmc9(l) = '1') AND bmc2(l) = '0' AND bmc8(7) = '0' AND bmc2(7) = '0')

OR

(bmc8(l) = '0' AND bmc2(l) = '1' AND (bmc2(0) = '1' OR bmcl(0) = '1' OR bmcl(l) = '1') AND bmc8(7) = '0' AND bmc2(7) = '0')

OR

(bmc8(l) = '0' AND bmc2(l) = '0' AND bmc8(7) = '1' AND (bmc9(8) = '1' OR bmc8(8) = '1' OR bmc9(7) = '1') AND bmc2(7) = '0')

OR

(bmc8(l) = '0' AND bmc2(l) = '0' AND bmc8(7) = '0' AND bmc2(7) = '1' AND (bmcl(7) = '1' OR bmc2(8) = '1' OR bmcl(8) = '1'))

OR

-- if exactly two corners are set and they are adjacent, then this is text not grayscale.

(bmc8(l) = '1' AND bmc2(l) = '1' AND bmc8(7) = '0' and bmc2(7)

0')

OR (bmc8(l) = '0' AND bmc2(l) = '1' AND bmc8(7) = '0' and bmc2(7)

= '1')

OR (bmc8(l) = '0' AND bmc2(l) = '0' AND bmc8(7) = '1' and bmc2(7)

OR (bmc8(l) = '1' AND bmc2(l) = '0' AND bmc8(7) = '1' and bmc2(7)

^■0' )

)

) ) ELSE

RightGrayDetected <= '0' WHEN (

— If certain pixels are black, then no grayscale enhancement. bmc5(0) = '1' OR bmc4(0) = '1' OR bmc3(0) = '1' OR bmc5(l) = '1' OR bmc4(l) = '1' OR bmc3(l) = '1' OR bmc5(2) = '1' OR bmc4(2) = '1' OR bmc3(2) = '1' OR bmc5(6) = '1' OR bmc4(6) = '1' OR bmc3(6) = '1' OR bmc5(7) = '1' OR bmc4(7) = '1' OR bmc3(7) = 'l¹ OR bmc5(8) = '1' OR bmc4(8) = '1' OR brac3(8) = '1' OR bmc8(3) = '1' OR bmc7(3) = '1' OR bmc6(3> = ' 1' OR bmc2(3) = ' 1' OR bmcl(3) = '1' OR bmcO(3) = '1' OR bmc8(4) = '1' OR bmc7(4) = '1' OR bmc6(4) = '1' OR bmc2(4) = '1' OR bmcl(4) = ^■ 1' OR bmc0(4) = '1' OR bmc8(5) = ' 1' OR bmc7(5) = '1' OR bmc6(5) = ' 1' OR bmc2(5) = ' 1' OR bmcl(5) = ' 1' OR bmc0(5) = '1' OR

(bmc4(4) = '0' AND — if middle pixel is white...

-- If just one corner is set, make sure that no adjacent pixels are set. If so this must be text¹

(bmc7(l) = '1' AND (bmc8(0) = ^■ 1' OR bmc7(0) = '1' OR bmc8(l) = '1') AND bmcl(l) = '0' AND bmc7(7) = '0' AND bmcl(7) = '0')

OR

(bmc7(l) = '0' AND bmcl(l) = ' 1' AND (bmcl(0) = ' 1' OR bmcO(O) = ' 1' OR bmcθ(l) = '1') AND bmc7(7) = '0' AND bmcl(7) = '0')

OR

(bmc7(l) = '0' AND bmcl(l) = '0' AND bmc7(7) = ' 1' AND (bmc8(8) = ' 1' OR bmc7(8) = '1' OR bmc8(7) = '1') AND bmcl(7) = '0')

OR

(bmc7(l) = '0' AND bmcl(l) = '0' AND bmc7(7) = '0' AND bmcl(7) = '1' AND (bmc0(7) = '1' OR bmcl(8) = '1' OR bmc0(8) = '1'))

OR

(bmc7(l) = ' 1' AND bmcl(l) = '1' AND bmc7(7) = '0' and bmcl(7)

= ' n 0 '')

OR (bmc7(l) = '0' AND bmcl(l) = ' 1' AND bmc7(7) = '0' and bmcl(7)

= ' 1 1 '")

OR

(bmc7(l) = '0' AND bmcl(l) = '0' AND bmc7(7) = ^■ 1' and bmcl(7)

= '1

OR (bmc7(l) = '1' AND bmcl(l) = '0' AND bmc7(7) = '1' and bmcl(7)

'0' ) ELSE

' l ' ;

— This has to count how many of the 5 pixels are set!

-- Counting is not easy, so I broke it into three stages:

The first stage combines two source pixels to counts of either 0, 1, or

The second stage takes those results to counts of 0-4 and the thrid stage adds the center pixel for counts of 0-5

LA2 <= bmc8(l) AND bmc8(7); — set if count is 2

LAI <= bmc8(l) XOR bmc8(7); — set if count is 1

LAO <= NOT bmc8(l) AND NOT bmc8(7);

LB2 <= bmc2(l) AND bmc2(7); — Set if count is 2

LB1 <= bmc2(l) XOR bmc2(7); — set if count is 1

LBO <= NOT bmc2(l) AND NOT bmc2(7);

LC4 <= LA2 AND LB2;

LC3 <= (LA2 AND LB1) OR (LAI AND LB2);

LC2 <= (LA2 AND LBO) OR (LB2 AND LAO) OR (LAI AND LB1);

LCI <= (LAI AND LBO) OR (LB1 AND LAO);

LCO <= LAO AND LBO;

LeftCount(O) <= LCO AND NOT bmc5(4);

LeftCount(l) <= (LCO AND bmc5(4)) OR (LCI AND NOT bmc5(4))

LeftCount(2) <= (LCI AND bmc5(4)) OR (LC2 AND NOT bmc5(4))

LeftCount(3) <= (LC2 AND bmc5(4)) OR (LC3 AND NOT bmc5(4))

LeftCount(4) <= (LC3 AND bmc5(4)) OR (LC4 AND NOT bmc5(4))

LeftCount(5) <= LC4 AND bmc5(4);

RA2 <= bmc7(l) AND bmc7(7); — set if count is 2

RA1 <= bmc7(l) XOR bmc7(7); — set if count is 1

RAO <= NOT bmc7(l) AND NOT bmc7(7);

RB2 <= bmcl(l) AND bmcl(7); — Set if count is 2

RBI <= bmcl(l) XOR bmcl(7); — set if count is 1 RBO <= NOT bmcl(l) AND NOT bmcl(7);

RC4 <= RA2 AND RB2 ;

RC3 <= (RA2 AND RBI) OR (RA1 AND RB2 ) ;

RC2 <= (RA2 AND RBO) OR (RB2 AND RAO) OR (RA1 AND RBI);

RC1 <= (RA1 AND RBO) OR (RBI AND RAO);

RCO <= RAO AND RBO;

RightCount (0) <= RCO AND NOT bmc4(4);

RightCount (1) <= (RCO AND bmc4(4)) OR (RC1 AND NOT bmc4(4))

RightCount (2) <= (RC1 AND bmc4(4)) OR (RC2 AND NOT bmc4(4))

RightCount (3) <= (RC2 AND bmc4(4)) OR (RC3 AND NOT bmc4(4))

RightCount (4) <= (RC3 AND bmc4(4)) OR (RC4 AND NOT bmc4(4))

RightCount (5) <= RC4 AND bmc4(4);

— A valid 1-bit grayscale case is when either left or right is detected, but

-- not if both sides have all 5 pixels black. iGrayDetected <= (RightGrayDetected AND NOT RightCount (0) AND NOT RightCount (5) ) OR (LeftGrayDetected AND NOT LeftCount(O) AND NOT LeftCount (5) ) ;

-- The symmetry only matters for the non-equal count pairs, but we always calculate it .

— Simply, the bit is set when the right count is greater than or equal to the left. iGsymmetry <= RightCount ( 5 ) OR

(NOT LeftCount (5) AND

(RightCount (4) OR

(NOT LeftCount (4) AND

(RightCount (3) OR

(NOT LeftCount (3) AND

(RightCount (2) OR

(NOT LeftCount (2) AND

RightCount (1) ))))))); — Outside of this module, the case numbers start at 220. Inside they start at 0.

— This table is the count of black pixels for each of the two center pixels.

— So if the left pixel count is 3 and the right count is 1, that is a case 7 (227) .

0 1 2 3 4 5

— 0 - 0 1 2 3 4

— ^• 1 0 5 6 7 8 9

— 2 1 6 10 11 12 13

-- 3 2 7 11 14 15 16

-- 4 3 8 12 15 17 18

-- 5 4 9 13 16 18 -

-- Cases 5, 10, 14, and 17 have the same count from each side, so they are easy. ιGcase(5) <= lGrayDetected AND LeftCount (1) AND RightCount ( 1) ; ιGcase(lO) <= lGrayDetected AND LeftCount (2) AND RightCount (2) ; ιGcase(14) <= lGrayDetected AND LeftCount (3) AND RightCount (3) ; ιGcase(17) <= lGrayDetected AND LeftCount (4) AND RightCount ( 4 ) ;

—The remaining cases have two possible counts that hit. ιGcase(O) <= lGrayDetected AND ( (LeftCount ( 1) AND RightCount (0) ) OR (LeftCount (0) AND RightCount ( 1 ) ιGcase(l) <= lGrayDetected AND (LeftCount (2) AND RightCount ( 0 ) ) OR (LeftCount (0) AND RightCount (2 ) ) ιGcase(2) <= lGrayDetected AND (LeftCount (3) AND RightCount ( 0 ) ) OR (LeftCount (0) AND RightCount (3) ) ιGcase(3) <= lGrayDetected AND (LeftCount (4) AND RightCount (0) ) OR (LeftCount (0) AND RightCount ( 4 ) ) ιGcase(4) <= lGrayDetected AND (LeftCount (5) AND RightCount (0) ) OR (LeftCount (0) AND RightCount ( 5 ) ) ιGcase(6) <= lGrayDetected AND (LeftCount (2) AND RightCount ( 1 ) ) OR (LeftCount (1) AND RightCount ( 2 ) ) ιGcase(7) <= lGrayDetected AND (LeftCount (3) AND RightCount ( 1 ) ) OR (LeftCount (1) AND RightCount (3 ) ) ιGcase(8) <= lGrayDetected AND ( (LeftCount (4) AND RightCount ( 1 ) ) OR (LeftCount (1) AND RightCount ( 4 ) ) ιGcase(9) <= lGrayDetected AND ( (LeftCount (5) AND RightCount ( 1 ) ) OR (LeftCount (1) AND RightCount ( 5 ) ) lGcase ( 11 ) <= lGrayDetected AND ( (LeftCount (3) AND RightCount (2 ) OR (LeftCount (2) AND RightCount (3) ) ) ; lGcase (12) <= lGrayDetected AND ( (LeftCount (4 ) AND RightCount (2 ) OR (LeftCount (2) AND RightCount ( 4 ) ) ) ; lGcase (13) <= lGrayDetected AND ( (LeftCount (5 ) AND RightCount (2 ) OR (LeftCount (2) AND RightCount (5) ) ) ; lGcase (15) <= lGrayDetected AND ( (LeftCount ( ) AND RightCount ( 3 ) OR (LeftCount (3) AND RightCount (4) ) ) ; lGcase (16) <= lGrayDetected AND ( (LeftCount ( 5 ) AND RightCount ( 3) OR (LeftCount (3) AND RightCount (5) ) ) ; lGcase (18) <= lGrayDetected AND ( (LeftCount (5) AND RightCount ( 4 ) OR (LeftCount (4; AND RightCount (5) ) ) ;

smglebitgray : PROCESS

BEGIN -- {

WAIT UNTIL elk' EVENT AND (elk =

GrayDetected <= lGrayDetected; xGsymmerry <= lGsymmetry; Gsymmetry <= xGsymmetry;

LatchCase: FOR n IN 0 to 18 GENERATE

Gcase(n) <= ιGcase(n); END GENERATE LatchCase;

Gcase (0 <= ιGcase(O) Gcase ( 1 <= ιGcase(l) Gcase (2 <= ιGcase(2) Gcase (3 <= ιGcase(3) Gcase (4 <= ιGcase(4) Gcase (5 <= ιGcase(5) Gcase ( 6 <= lGcase ( 6) Gcase (7 <= ιGcase(7) Gcase (8 <= ιGcase(8) Gcase (9) <= lGcase (9) ; Gcase (10 <= ιGcase(lO) Gcase (11 <= ιGcase(ll) Gcase (12 <= ιGcase(12) Gcase (13 <= ιGcase(13) Gcase (14 <= ιGcase(14) Gcase (15 <= ιGcase(15) Gcase ( 16 <= ιGcase(16) Gcase (17 <= ιGcase(17) Gcase(18 <= ιGcase(18)

END PROCESS singlebitgray; — }

END structural;

The grayscale enhancement technique described in the listing above has been applied to a 141 dpi line screen with 144 shades of gray printed on a standard 600 dpi laser printer. The quality achieved approaches magazine quality and matches or surpasses other multi-bit image enhancement techniques that require more memory, software modifications and increases in processing requirements — all of which result in poorer performance and/or increases in controller costs. The enhancement method of this invention works with existing printer software and may be integrated with text and line-art enhancement to produce a total package for producing quality text, line-art and photo images.

Although specific features of this invention are shown in some drawings and not others, this is for convenience only as each feature may be combined with any or all of the other features in accordance with the invention.

Other embodiments will occur to those skilled in the art and are within the following claims:

Claims

What is claimed is:

1. A grayscale enhancement method comprising: analyzing the state of selected pixels surrounding a subject pixel in a bit map; determining the number of selected pixels including the subject pixel that are black; and printing, at a location corresponding to the location of the subject pixel in the bit map, a fractional pixel related in size to at least the computed number of pixels which are black.

2. The method of claim 1 in which the step of analyzing includes generating a window which moves about the bit map.

3. The method of claim 2 in which the window is nine by ten pixels in area.

4. The method of claim 2 in which the selected surrounding pixels are located at a predetermined set of vectors with respect to the subject pixel within the window.

5. The method of claim 4 in which the predetermined set of vectors each have a length n.

6. The method of claim 5 in which n is 3^/2 pixels.

7. The method of claim 4 in which each vector extends to a different corner of the window.

8. The method of claim 1 in which m pixels are selected as the surrounding pixels and printing includes driving a printer modulator to print a pixel of a size approximately x/(m+ l) where x is the number of black pixels present in the set including the m surrounding pixels and the subject pixel.

9. The method of claim 10 in which m equals 4.

10. A grayscale enhancement method comprising: analyzing the state of a plurality of pixels in a portion of a bit map; generating a number of fractional size pixels for each black pixel of the plurality; and printing the fractional size pixels.

11. The method of claim 10 in which the plurality contains m pixels and each fractional size pixel has a size 1/m.

12. The method of claim 11 in which the number of fractional size pixels is m.

13. A grayscale enhancement system comprising: means for analyzing the state of selected pixels surrounding a subject pixel in a bit map; means for determining, from the selected pixels and the subject pixel, the number of black pixels; and means for printing, at a location corresponding to the location of the subject pixel in the bit map, a fractional pixel related in size to at least the computed number of pixels which are black.

14. The system of claim 13 in which the means for analyzing includes means for generating a window which moves about the bit map.

15. The system of claim 14 in which the window is nine by ten pixels in area.

16. The system of claim 14 in which the means for analyzing further includes means for selecting surrounding pixels that are located at a predetermined set of vectors with respect to the subject pixel within the window.

17. The system of claim 16 in which the predetermined set of vectors each have a length n.

18. The system of claim 17 in which n is 3^/2 pixels.

19. The system of claim 16 in which each vector extends to a different corner of the window.

20. The system of claim 13 in which printing includes driving a printer modulator to print a pixel of a size approximately x/(m+ l) where x is the number of black pixels present in the set including m surrounding pixels and the subject pixel.

21. The system of claim 20 in which m equals 4.

22. A grayscale enhancement system comprising: means for analyzing the state of a plurality of pixels in a portion of a bit map; means for generating a number of fractional size pixels for each black pixel of tl lurality; and means for printing the fractional size pixels.

23. The system of claim 22 in which the plurality contains m pixels and each fractional size pixel has a size 1/m.

24. The system of claim 23 in which the number of fractional size pixels is m.