JP3611697B2 - Multilevel image binarization method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a multi-value image binarization method for converting a multi-value image into a binary image, and in particular, a multi-value image binary for converting a multi-value image into a binary image having a higher density than the multi-value image. It relates to the conversion method.
[0002]
[Prior art]
For example, in a digital copying machine or the like that outputs a multi-value image read by a scanner device with a plotter device corresponding to the binary image, the multi-value image is a binary image composed of binary pixels of white pixels and black pixels. It needs to be converted to an image. In the binarization processing, which level of the range that the density value of each pixel constituting the multi-valued image can take is set as a threshold value as a white pixel or a black pixel, the image quality of the binarized image is greatly increased. Affect.
[0003]
Therefore, as can be seen in “Threshold setting method for binarization of image density” described in Japanese Patent Application No. 56-46187, binary is used according to the connectivity of density levels of pixels of a 3 × 3 matrix. In the binarization of a multi-valued image, such as a method for setting a threshold value, a proposal has been made so that the features of the target image can be accurately reproduced.
[0004]
[Problems to be solved by the invention]
However, the method disclosed in the above Japanese Patent Publication No. 56-46187 requires a page memory for holding multi-valued image data in order to obtain a connectivity histogram, and an apparatus for realizing the method is expensive. Become.
[0005]
In recent years, the resolution of an input device such as an image scanner for reading a multi-valued image is mainly 300 to 400 dpi, but the resolution of an output device such as a plotter device for outputting a binary image is 600 to 600. As the density has increased to 1200 dpi, it is necessary to convert a multi-valued image into a binary image having a higher density than the pixel density of the multi-valued image. There was a problem that it was not possible.
[0006]
The present invention has been made in view of such circumstances, and an object of the present invention is to provide a multi-valued image binarization method that does not require a page memory and can convert a multi-valued image into an optimal binary image while increasing the density. And
[0007]
[Means for Solving the Problems]
In order to achieve the above object, the multi-value image binarization method according to claim 1 uses each pixel of a multi-value image composed of pixels having multi-value density values as a target pixel, and performs main scan on the target pixel. The multi-valued image is converted into (S × T) binary pixels by dividing in the direction S (S is a natural number) times and the sub-scanning direction T (T is a natural number) times, thereby converting the multi-valued image into a high-density binary image. In the multi-value image binarization method for conversion, the density value of the pixel of interest is simply converted to (S × T + 1) gradation with (S × T) fixed threshold values. Multi-value The gradation output value obtained as a result of the conversion process is the number of black pixels among the (S × T) binary pixels converted from the pixel of interest, and the pixel of interest is the number of gradation output values. To S × T binary pixels including black pixels And before Whether or not the pixel of interest is an edge pixel is determined from the distribution of density values of (M × N) peripheral pixels (M and N are natural numbers) of the pixel of interest. Meanwhile, the arrangement position of the black pixels included in the (S × T) binary pixels converted from the target pixel is determined according to the distribution state of the density values of the (M × N) peripheral pixels. When the target pixel is not an edge pixel, the arrangement position of the black pixel included in the (S × T) binary pixels converted from the target pixel is determined according to a predetermined priority. It is characterized by Many Value image binarization method.
[0008]
3. The multi-value image binarization method according to claim 2, wherein each pixel of the multi-value image composed of pixels having multi-value density values is set as a target pixel, and the target pixel is set in the main scanning direction S (S is a natural number). Multi-valued image binary that converts the multi-valued image into a high-density binary image by dividing it into double and sub-scanning direction T (T is a natural number) times and converting it into (S × T) binary pixels In the conversion method, the gradation output value obtained as a result of performing the (S × T + 1) gradation multi-value error diffusion processing on the density value of the pixel of interest is converted from the pixel of interest (S × T). As the number of black pixels among the binary pixels, the pixel of interest is converted into (S × T) binary pixels including the black pixels corresponding to the gradation output values. And before Whether or not the pixel of interest is an edge pixel is determined from the distribution of density values of (M × N) peripheral pixels (M and N are natural numbers) of the pixel of interest. Meanwhile, the arrangement position of the black pixels included in the (S × T) binary pixels converted from the target pixel is determined according to the distribution state of the density values of the (M × N) peripheral pixels. When the target pixel is not an edge pixel, the arrangement position of the black pixel included in the (S × T) binary pixels converted from the target pixel is determined according to a predetermined priority. It is characterized by Many Value image binarization method.
[0009]
4. The multi-value image binarization method according to claim 3, wherein each pixel of the multi-value image composed of pixels having multi-value density values is set as a target pixel, and the target pixel is set in the main scanning direction S (S is a natural number). Multi-valued image binary that converts the multi-valued image into a high-density binary image by dividing it into double and sub-scanning direction T (T is a natural number) times and converting it into (S × T) binary pixels In the conversion method, the gradation output value obtained as a result of performing the (S × T + 1) gradation multi-value dither processing on the density value of the target pixel is converted from the target pixel (S × T). As the number of black pixels among the binary pixels, the target pixel is converted into (S × T) binary pixels including only the gradation output value black pixels. And before Whether or not the pixel of interest is an edge pixel is determined from the distribution of density values of (M × N) peripheral pixels (M and N are natural numbers) of the pixel of interest. Meanwhile, the arrangement position of the black pixels included in the (S × T) binary pixels converted from the target pixel is determined according to the distribution state of the density values of the (M × N) peripheral pixels. When the target pixel is not an edge pixel, the arrangement position of the black pixel included in the (S × T) binary pixels converted from the target pixel is determined according to a predetermined priority. It is characterized by Many Value image binarization method.
[0016]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
[0017]
FIG. 1 is a diagram showing a block configuration of an image processing apparatus to which a multi-value image binarization method according to an embodiment of the present invention is applied.
[0018]
In the figure, a scanner unit 1 includes a photoelectric conversion device and an A / D conversion device, reads a document, converts it into multi-value image data, and outputs it. The binarizing unit 2 converts the multi-value image data from the scanner unit 1 into binary image data having a density that is S times the main scanning direction and T times the sub-scanning direction. Note that S and T are integers of 1 or more, that is, natural numbers. In this embodiment, as described later, when S = T = 2, S = T = 3, or S = T = 4 Will be described as an example. The printer 3 records and outputs the binary image from the binarizing unit 2 at a density that is S times the scanning density of the scanner unit 1 and T times the sub scanning direction. The control unit 4 is a microcomputer that controls the above-described units.
[0019]
The multi-value image binarization method according to the present invention performed in the binarization unit 2 with the above configuration will be described separately for the first to eighth embodiments.
[0020]
First, in the first embodiment, the binarizing unit 2 sequentially inputs each pixel having multi-value density values constituting the multi-value image data input from the scanner unit 1 as a target pixel, and the target pixel. A simple multi-value conversion process is performed to convert the density value of the pixel to (S × T + 1) gradation with (S × T) fixed threshold values, and the resulting value from 0 to the value (S × T) The gradation output value in the range up to T) is the number of black pixels of the (S × T) binary pixels converted from the pixel of interest, and the pixel of interest is the black pixel corresponding to the number of gradation output values. Is converted into (S × T) binary pixels including the binary value, and the multi-valued image from the scanner unit 1 is densified by S times in the main scanning direction and T times in the sub-scanning direction. Output an image.
[0021]
Specifically, when S = T = 2, the target pixel X located at the center of the 3 × 3 peripheral pixels A, B, C, D, E, G, and H shown in FIG. (2 × 2) = 4 binary pixels a, b, c, and d are converted, and the number of black pixels in the (2 × 2) binary pixels and the (2 × 2) There are 16 combinations of black pixel arrangement positions in the matrix as shown in FIGS. (1) in the figure is when the number of black pixels is 0, (2) to (5) in the figure are when the number of black pixels is one, and (6) to (11) in FIG. In this case, FIGS. 12 to 15 show the case where the number of black pixels is three, and FIG. 16 shows the case where the number of black pixels is four. In the first embodiment, (2 X2) One of the patterns shown in FIGS. 4 (1) to (16) is selected corresponding to the number of black pixels in the binary pixels.
[0022]
The conversion of the density value of the target pixel into the number of black pixels in the (2 × 2) binary pixel is set to f (x, y) as the density value of the target pixel, and the density value f (x, y) is 4 The pixel of interest according to which one of the five density ranges divided by two threshold density values (th1, th2, th3, th4: where th4>th3>th2> th1) belongs This is done by determining the number of black pixels contained in (2 × 2) = 4 binary pixels shown in FIG.
[0023]
That is, the density value f (x, y) of the target pixel X is
If f (x, y) ≧ th4, the number of black pixels = 4
If th4> f (x, y) ≧ th3, the number of black pixels = 3
If th3> f (x, y) ≧ th2, the number of black pixels = 2
If th2> f (x, y) ≧ thl, the number of black pixels = l
If th1> f (x, y), the number of black pixels = 0
As described above, the number of black pixels included in the (2 × 2) binary pixel is determined as one of 0 to 4 according to the five-level density range to which the density value of the target pixel X belongs, and the determination is made. Depending on the number of black pixels, one of the patterns shown in FIGS. 4 (1) to (16) is selected.
[0024]
As described above, in the multi-value image binarization method of the first embodiment, the multi-value from the scanner unit 1 is obtained by a process that does not require a page memory for converting the target pixel into (S × T) binary pixels. The image can be converted into a high-density binary image and output to the printer 3. Further, the gradation output value obtained by performing the simple multi-value processing on the density value of the target pixel is set as the number of black pixels in the (S × T) binary pixels converted from the target pixel, thereby obtaining a binary value. The converting unit 2 can output to the printer 3 a high-density binary image suitable for a multi-valued image obtained by reading a character document with the scanner unit 1, and the printer 3 can record a character image with good image quality. .
[0025]
Next, a second embodiment will be described. In the second embodiment, the binarization unit 2 sequentially inputs each pixel having multi-value density values constituting the multi-value image data input from the scanner unit 1 as a target pixel, and the target pixel A multi-value error diffusion process of (S × T + 1) gradation is performed on the density value, and the gradation output value in the range from value 0 to value (S × T) obtained as a result is obtained as the target pixel. The number of black pixels in the (S × T) binary pixels to be converted from is converted into (S × T) binary pixels including the black pixel corresponding to the gradation output value. The multi-valued image from the scanner unit 1 is output as a binary image that is densified S times in the main scanning direction and T times in the sub-scanning direction.
[0026]
Specifically, in the case of S = T = 2, when multi-level error diffusion processing of 5 gradations is performed on a target pixel of 64 gradations (density value 0 to 63), 3 is shown in FIG. The target pixel X located at the center of the × 3 peripheral pixels A, B, C, D, E, G, H is (2 × 2) = 4 binary pixels a, b, c, And the number of black pixels (0 to 4) corresponding to the five gradations of the multilevel error diffusion processing in the (2 × 2) binary pixel, and the (2 × 2) matrix There are 16 combinations of black pixel arrangement positions as shown in FIGS. (1) in the figure is when the number of black pixels is 0, (2) to (5) in the figure are when the number of black pixels is one, and (6) to (11) in FIG. In this case, FIGS. 12 to 15 show a case where the number of black pixels is three, and FIG. 16 shows a case where the number of black pixels is four. In the second embodiment, (2 X2) One of the patterns shown in FIGS. 4 (1) to (16) is selected corresponding to the number of black pixels in the binary pixels.
[0027]
Conversion of the density value of 64 gradations of the target pixel into the number of black pixels in the (2 × 2) binary pixel is set to f (x, y) as the input density value of the target pixel, and the input density value f (x , Y) and the density value f ′ (x, y), which is the sum of the error diffused from the predetermined peripheral pixel of the target pixel to the target pixel, is the four threshold density values (th1, th2, th3, th4). However, it is shown in FIG. 3 that is converted from the target pixel X according to which one of the predetermined five density ranges divided by th4>th3>th2> th1). ) = By determining the number of black pixels included in the four binary pixels to any one of 0 to 4.
[0028]
That is, the density value of the pixel of interest X is f ′ (x, y) as described above, and the error of the peripheral pixel located at (xi, y−j) with respect to the position (x, y) of the pixel of interest. When the value is e (x−i, y−j), the diffusion coefficient is μ (i, j), and the error value of the pixel of interest is e (x, y),
f ′ (x, y) = f (x, y) + (1 / Σμ (i, j)) × Σμ (i, j) e (x−i, y−j)
F ′ (x, y) obtained by
If f ′ (x, y) ≧ th4, the number of black pixels = 4: e (x, y) = f ′ (x, y) −63
If th4> f ′ (x, y) ≧ th3, the number of black pixels = 3: e (x, y) = f ′ (x, y) −47
If th3> f ′ (x, y) ≧ th2, the number of black pixels = 2: e (x, y) = f ′ (x, y) −31
If th2> f ′ (x, y) ≧ thl, the number of black pixels = 1: e (x, y) = f ′ (x, y) −15
If th1> f ′ (x, y), the number of black pixels = 0: e (x, y) = f ′ (x, y)
As described above, the number of black pixels included in the (2 × 2) binary pixel is determined to be any one of 0 to 4 according to the five-level density range to which the density value of the target pixel X belongs, and the determination is made. Depending on the number of black pixels, one of the patterns shown in FIGS. 4 (1) to (16) is selected. At the same time, an error value e (x, y) for the current pixel of interest is calculated so that multi-value error diffusion processing can be performed in the neighboring pixels that will be processed as the pixel of interest in the vicinity of the pixel of interest at the same time. . The error value e (x, y) is a multi-value representing one pixel in a multi-valued image with 5 gradations expressed as 0 to 4 black pixels included in (2 × 2) = 4 pixels. Since it is in error diffusion processing, it is calculated by a mathematical formula corresponding to the number of black pixels in the (2 × 2) binary image converted from the target pixel. Note that the multi-level error diffusion process itself, which is an application of the error diffusion process, is a known technique, and a detailed description thereof will be omitted.
[0029]
As described above, in the multi-value image binarization method according to the second embodiment, the multi-value from the scanner unit 1 is obtained by a process that does not require a page memory for converting the target pixel into (S × T) binary pixels. The image can be converted into a high-density binary image and output to the printer 3. In addition, among the (S × T) binary pixels obtained by converting the gradation output value obtained by performing the multi-value error diffusion processing of (S × T + 1) gradations on the density value of the target pixel from the target pixel. The binarization unit 2 can output a high-density binary image suitable for a multi-valued image obtained by reading a photographic document with the scanner unit 1 to the printer 3. 3 can record a photographic image of good quality.
[0030]
Next, a third embodiment will be described. In the third embodiment, the binarization unit 2 sequentially inputs each pixel having multi-value density values constituting the multi-value image data input from the scanner unit 1 as a target pixel, and the target pixel (S × T + 1) gradation multi-value dither processing is performed on the density value, and the gradation output value in the range from the value 0 to the value (S × T) obtained as a result is obtained from the target pixel. The number of black pixels in the (S × T) binary pixels to be converted is set, and the pixel of interest is converted into (S × T) binary pixels including black pixels corresponding to the gradation output values. By outputting, the multi-valued image from the scanner unit 1 is output as a binary image having a high density of S times in the main scanning direction and T times in the sub-scanning direction.
[0031]
Specifically, in the case of S = T = 2 and when multi-level dither processing of 5 gradations is performed on a target pixel of 64 gradations (density value 0 to 63), the 3 × shown in FIG. The target pixel X located at the center of the three peripheral pixels A, B, C, D, E, G, and H is (2 × 2) = four binary pixels a, b, c, and , D, but the number of black pixels (0 to 4) corresponding to the five gradations of the multi-value dither processing in the (2 × 2) binary pixels, and the black pixels in the (2 × 2) matrix There are 16 combinations of arrangement positions in FIG. 4 (1) to (16). (1) in the figure is when the number of black pixels is 0, (2) to (5) in the figure are when the number of black pixels is one, and (6) to (11) in FIG. In this case, FIGS. 12 to 15 show the case where the number of black pixels is three, and FIG. 16 shows the case where the number of black pixels is four. In the third embodiment, (2 X2) One of the patterns shown in FIGS. 4 (1) to (16) is selected corresponding to the number of black pixels in the binary pixels.
[0032]
The conversion of the density value of 64 gradations of the target pixel into the number of black pixels in the (2 × 2) binary pixel is set to f (x, y) as the density value of the target pixel. ) Is divided into four threshold density values (th (i, j), th (i, j) +15, th (i, j) +31, th (i, j) +47). The number of black pixels included in (2 × 2) = 4 binary pixels shown in FIG. 3 converted from the target pixel X according to which density range belongs to 0 to 4 This is done by deciding on one of these.
[0033]
Note that th (i, j) is a fluctuation threshold value, and is defined by a 4 × 4 fluctuation threshold value matrix as shown in FIG. For example, th (1, 1) = 12, th (4, 4) = 14. In the 4 × 4 variation threshold value matrix shown in FIG. 5, which is virtually tiled on the multi-value image data from the scanner unit 1 input to the binarization unit 2, the target pixel is the main pixel. Depending on which position (i, j) in the scanning direction (horizontal direction in the figure) and sub-scanning direction (vertical direction in the figure), the variation threshold th (i, j) for the pixel of interest Is decided.
[0034]
That is, the density value f (x, y) of the target pixel X is
If f (x, y) ≧ th (i, j) +47, the number of black pixels = 4
If th (i, j) +47> f (x, y) ≧ th (i, j) +31, the number of black pixels = 3
If th (i, j) +31> f (x, y) ≧ th (i, j) +15, the number of black pixels = 2
If th (i, j) +15> f (x, y) ≧ th (i, j), the number of black pixels = l
If th (i, j)> f (x, y), the number of black pixels = 0
As described above, the number of black pixels included in the (2 × 2) binary pixel is determined to be any one of 0 to 4 according to the five-level density range to which the density value of the target pixel X belongs, and the determination is made. Depending on the number of black pixels, one of the patterns shown in FIGS. 4 (1) to (16) is selected. Note that multi-value dither processing itself, which is an application of dither processing, is a known technique, and thus detailed description thereof is omitted.
[0035]
As described above, in the multi-valued image binarization method of the third embodiment, the multi-value from the scanner unit 1 is obtained by a process that does not require a page memory to convert the pixel of interest into (S × T) binary pixels. The image can be converted into a high-density binary image and output to the printer 3. In addition, the gradation output value obtained by performing the multi-value dither processing of (S × T + 1) gradation on the density value of the target pixel is converted from the target pixel in (S × T) binary pixels. By setting the number of black pixels, the binarizing unit 2 can output to the printer 3 a high-density binary image suitable for a multi-valued image obtained by reading a photographic document with the scanner unit 1. Can record photographic images of good quality.
[0036]
Next, a fourth embodiment will be described. The fourth embodiment is a modification of the multilevel image binarization method of the first to third embodiments.
[0037]
In the first to third embodiments described above, the gradation output value of the target pixel is obtained by simple multi-value processing, multi-value error diffusion processing, or multi-value dither processing, and converted from the target pixel (S × T ) The arrangement position of the black pixels in the (S × T) binary image is not particularly limited simply by setting the number of black pixels in the binary image to the obtained gradation output value.
[0038]
Accordingly, the fourth embodiment is intended to optimize the arrangement of the number of black pixels obtained in the first to third embodiments in the (S × T) binary image.
[0039]
That is, in the fourth embodiment, the binarization unit 2 includes each pixel of the multi-valued image data input from the scanner unit 1 as a target pixel, and includes the number of black pixels corresponding to the density value of the target pixel. As described above, the arrangement of black pixels in the (S × T) binary pixel converted from the pixel of interest is the density value of (M × L) peripheral pixels (M and L are natural numbers) around the pixel of interest. It is determined according to the distribution situation of.
[0040]
Specifically, taking the case of S = T = 2 and M = L = 3 as an example, the arrangement of black pixels in the (2 × 2) binary pixels converted from the target pixel is as described above. Depending on the number of black pixels included, the patterns shown in FIGS. In the following description, it is assumed that the (3 × 3) peripheral pixels of the target pixel X are pixels A to H arranged as shown in FIG.
[0041]
When the number of black pixels is 0, there is only one pattern shown in FIG. 4 (1), so there is no room for selection, and (3 × 3) in FIG. ) Is selected. Further, when the number of black pixels is four, there is only one pattern shown in FIG. 4 (16), so there is no room for selection, and FIG. 4 does not depend on the distribution state of each density value of (3 × 3) peripheral pixels. The pattern shown in (16) is selected.
[0042]
If the number of black pixels is 3, one of the patterns shown in FIGS. 4 (12) to (15) can be selected. In that case, assuming that the (2 × 2) binary pixel matrix shown in FIG. 3 is applied to the frame of the target pixel X in FIG. 2, the density values of the peripheral pixels A, C, F, and H are compared, and Among them, a binary pixel adjacent to a peripheral pixel having the smallest density value and a thin one is a white pixel, and the remaining three are black pixels. That is, when the peripheral pixel A has the lowest density value, the pattern of FIG. 4 (12) is selected, and when the peripheral pixel C has the lowest density value, the pattern of FIG. 4 (13) is selected. When F has the smallest density value, the pattern of FIG. 4 (15) is selected, and when the peripheral pixel H has the smallest density value, the pattern of FIG. 4 (14) is selected.
[0043]
If the number of black pixels is two, any one of the patterns of FIGS. 4 (6) to (11) can be selected. In that case, assuming that the (2 × 2) binary pixel matrix shown in FIG. 3 is applied to the frame of the target pixel X in FIG. 2, the total density values of the two neighboring pixels located on the diagonal are compared. Two binary pixels along the diagonal with the largest total density value are black pixels. That is, the total density value is compared with four sets of A + H, B + G, C + F, and D + E. If A + H is maximum, the pattern of FIG. 4 (10) is selected, and if B + G is maximum, FIG. 4 (8) or (9 4), the pattern of FIG. 4 (11) is selected if C + F is maximum, and the pattern of FIG. 4 (6) or (7) is selected if D + E is maximum.
[0044]
If the number of black pixels is one, any of the patterns shown in FIGS. 4 (2) to (5) can be selected. In that case, assuming that the (2 × 2) binary pixel matrix shown in FIG. 3 is applied to the frame of the target pixel X in FIG. 2, the density values of the peripheral pixels A, C, F, and H are compared. Among them, a binary pixel adjacent to a peripheral pixel having the largest density value is a black pixel, and the remaining three are white pixels. That is, when the peripheral pixel A has the highest density value, the pattern shown in FIG. 4 (2) is selected, and when the peripheral pixel C has the highest density value, the pattern shown in FIG. 4 (3) is selected. When F has the highest density value, the pattern of FIG. 4 (5) is selected, and when the peripheral pixel H has the highest density value, the pattern of FIG. 4 (4) is selected.
[0045]
As described above, in the multilevel image binarization method of the fourth embodiment, the arrangement position in the (S × T) binary pixel matrix of the number of black pixels determined by the first to third embodiments is determined as the target pixel. The black pixels included in the (S × T) binary pixel matrix converted from the target pixel are preferentially arranged on the high density side of the peripheral pixels in accordance with the distribution of density values of the (M × N) peripheral pixels. Therefore, the binarizing unit 2 can convert the edges in the multi-valued image input from the scanner unit 1 into a fine high-density binary image and output it to the printer 3. In the printer 3, It is possible to output a high-density binary image that clearly reproduces the characters in the multi-valued image and the outline of the photograph.
[0046]
Next, a fifth embodiment will be described. The fifth embodiment is a modified example different from the fourth embodiment of the multilevel image binarization method of the first to third embodiments.
[0047]
In the first to third embodiments described above, the gradation output value of the target pixel is obtained by simple multi-value processing, multi-value error diffusion processing, or multi-value dither processing, and converted from the target pixel (S × T ) The arrangement position of the black pixels in the (S × T) binary image is not particularly limited simply by setting the number of black pixels in the binary image to the obtained gradation output value.
[0048]
Therefore, the fifth embodiment is intended to optimize the arrangement of the number of black pixels obtained in the first to third embodiments in the (S × T) binary image.
[0049]
That is, in the fifth embodiment, the binarization unit 2 includes each pixel of the multi-valued image data input from the scanner unit 1 as a target pixel, and includes the number of black pixels corresponding to the density value of the target pixel. As described above, the arrangement of the black pixels in the (S × T) binary pixel converted from the target pixel is determined according to a predetermined priority.
[0050]
Specifically, taking the case of S = T = 2 as an example, the arrangement of black pixels in the (2 × 2) binary pixels converted from the target pixel is the number of black pixels included as described above. Accordingly, the patterns shown in FIGS. 4 (1) to (16) are obtained.
[0051]
When there are a plurality of patterns that can be selected according to the number of black pixels included, for example, when the number of black pixels included is one, four patterns shown in FIGS. 4 (2) to (5) can be selected. When the number of black pixels included is two, the six patterns shown in FIGS. 4 (6) to (11) can be selected. When the number of black pixels included is three, the patterns shown in FIGS. The four patterns (15) can be selected.
[0052]
Therefore, in the fifth embodiment, priorities shown in the priority matrix shown in FIG. 6 are assigned to (2 × 2) binary pixels. If the priority matrix shown in FIG. 6 and the (2 × 2) binary pixel matrix shown in FIG. 3 are superposed, the priority order is higher in the order of binary pixels b, d, a, and c. 2) Black pixels included in binary pixels are arranged in the order of binary pixels b, d, a, and c.
[0053]
Therefore, when the number of black pixels included in the (2 × 2) binary pixel is one, the pattern shown in FIG. 4C is selected. When the number of included black pixels is two, the pattern shown in FIG. When a pattern is selected and three black pixels are included, the pattern shown in FIG. 4 (15) is selected. If the number of black pixels included is zero, only the pattern shown in FIG. 4A can be selected, so the pattern shown in FIG. 4A is selected. The same applies to the case where four black pixels are included, and only the pattern of FIG. 4 (16) can be selected, so the pattern of FIG. 4 (16) is selected.
[0054]
As described above, in the multilevel image binarization method of the fifth embodiment, the arrangement positions in the (S × T) binary pixel matrix of the number of black pixels determined by the first to third embodiments are determined in advance. Therefore, the black pixels included in the (S × T) binary pixels converted from the target pixel can be regularly arranged, and the binarization unit 2 includes the scanner unit 1. A multi-valued image obtained by reading a photograph input from the image can be converted into a high-density binary image having a regular gradation and a smooth gradation, and can be output to the printer 3. In the printer 3, the gradation is regular. A smooth high-density binary image can be output.
[0055]
Next, a sixth embodiment will be described. The sixth embodiment is a modification of the multi-value image binarization method of the fifth embodiment.
[0056]
That is, in the sixth embodiment, in the fifth embodiment, the binarization unit 2 referred to when determining the arrangement of black pixels in the (S × T) binary pixels converted from the target pixel. The priority of (S × T) binary pixels is set to a priority at which the converted high-density binary image becomes a vertical stripe image.
[0057]
Specifically, taking the case of S = T = 2 as an example, the priority shown in the priority matrix shown in FIG. 6 is overlapped with (2 × 2) binary pixels to correspond. The priority matrix shown in FIG. 6 is also referred to in the fifth embodiment. Actually, in the priority matrix shown in FIG. 6, the converted high-density binary image is a priority that becomes a vertical stripe image. The degree is set.
[0058]
That is, if the priority matrix shown in FIG. 6 and the (2 × 2) binary pixel matrix shown in FIG. 3 are overlapped, the first binary pixel b and the second binary pixel d having the priority order. Are continuous in the vertical direction (sub-scanning direction), and the number of black pixels included in the (2 × 2) binary pixel converted from the target pixel is two, the two black pixels are , Will be lined up vertically. Therefore, the converted high-density binary image as a whole is a vertical streak-like image extending in the sub-scanning direction every other pixel in the main scanning direction (horizontal direction).
[0059]
Therefore, it is less susceptible to jitter caused by uneven feeding of the recording paper in the sub-scanning direction in the printer 3 that may occur when the high-density binary image output from the binarization unit 2 is recorded on the recording paper by the printer 3. A high-density binary image can be output in a high quality state. Even if the high-density binary image after conversion is an image with vertical stripes, the black pixel converted from the target pixel by the multilevel image binarization method of the first to third embodiments described above. Since the number is optimally determined according to the density value in the multi-valued image before conversion, black pixels are vertically aligned within the small area of binary pixels (S × T) converted from the target pixel. Even if they are arranged, in the high-density binary image after conversion, the vertical stripes that are visibly visible are not generated and the image quality is not deteriorated.
[0060]
Next, a seventh embodiment will be described. The seventh embodiment is a modification of the multi-value image binarization method of the sixth embodiment.
[0061]
That is, in the sixth embodiment, by setting the priority of the (S × T) binary pixel converted from the target pixel to the priority at which the converted high-density binary image becomes a vertical stripe image, The printer 3 can be less susceptible to jitter caused by uneven feeding of the recording paper in the sub-scanning direction. However, when the recording density of the printer 3 is high, for example, in the case of the sixth embodiment, the resolution of the printer 3 is about 200 dpi. Thus, in the case of the seventh embodiment, when the resolution is 400 dpi or more, as in the sixth embodiment, every other pixel in the main scanning direction (lateral direction) as the entire high-density binary image after conversion, Even if the image can be a vertical streak-like image extending in the sub-scanning direction, the vertical streak width is too narrow at the time of recording by the printer 3, and there is a risk that the image will be blurred.
[0062]
Therefore, in the seventh embodiment, when the case of S = T = 2 is taken as an example, the binarizing unit 2 pays attention to the odd-numbered pixels in the main scanning direction of the multivalued image data input from the scanner unit 1. When processing as a pixel, the priority matrix shown in FIG. 6 is adopted, and when processing an even-numbered image as a target pixel, a priority matrix shown in FIG. 7 and symmetrical with respect to the main scanning direction is shown in FIG. Is adopted. Of course, the priority matrix shown in FIG. 6 may be adopted evenly, and the priority matrix shown in FIG. 7 may be adopted odd.
[0063]
As shown in FIG. 3 and FIG. 3, the priority matrixes referred to for the pixels adjacent to each other in the main scanning direction are symmetrical to the left and right (main scanning direction). X2) If the binary pixel matrix is overlapped, the priority order between the pixel to which the priority matrix of FIG. 6 is applied and the pixel to which the priority matrix of FIG. The first binary pixel b and the second binary pixel d, and the first binary pixel a and the second binary pixel c, which are prioritized, are adjacent to each other. Doubles in thickness. Therefore, even if the recording resolution of the printer 3 is high, it is possible to obtain a good recorded image in which the vertical stripes are difficult to fade.
[0064]
Next, an eighth embodiment will be described. The eighth embodiment is a modification of the multi-value image binarization method of the fifth embodiment described above.
[0065]
That is, in the eighth embodiment, in the fifth embodiment, the binarization unit 2 is referred to when determining the arrangement of black pixels in the (S × T) binary pixels converted from the target pixel. The priority of (S × T) binary pixels is set to a priority at which the converted high-density binary image becomes a halftone image.
[0066]
Specifically, taking the case of S = T = 4 as an example, the priority shown in the priority matrix shown in FIG. 8 is overlapped with (4 × 4) binary pixels to correspond. In the priority matrix shown in FIG. 8, priorities are set such that the converted high-density binary image is a halftone image.
[0067]
That is, if the priority matrix shown in FIG. 8 and the (4 × 4) binary pixel matrix converted from the target pixel are overlapped, the first to third included in the (4 × 4) binary pixel matrix. If the number of black pixels determined by the multi-valued image binarization method of the embodiment is up to 4, the black pixels are sequentially arranged in the 2 × 2 pixels in the center of the binary pixel matrix in the priority order shown in FIG. Even in the case of four or more pixels, the black pixels are arranged as close to the center of the (4 × 4) binary pixel matrix as possible.
[0068]
In this way, by setting the priority of (S × T) binary pixels to a priority at which black pixels in the (S × T) binary pixel matrix are arranged in the center of the matrix as much as possible, Since the binary density image becomes a halftone dot image, the binarizing unit 2 converts the multi-valued image input from the scanner unit 1 into a high-density binary image having a regular gradation and a smooth gradation. The printer 3 can output a high-density binary image with regular and smooth gradation of photographs in a multi-valued image.
[0069]
Next, a ninth embodiment will be described. The ninth embodiment is a modification of the multilevel image binarization method of the first to third embodiments described above.
[0070]
In the first to third embodiments described above, the gradation output value of the pixel of interest is obtained by simple multi-value processing, multi-value error diffusion processing, or multi-value dither processing, and converted from the pixel of interest (S × T ) The arrangement position of the black pixels in the (S × T) binary image is not particularly limited simply by setting the number of black pixels in the binary image to the obtained gradation output value.
[0071]
Therefore, this ninth embodiment is intended to optimize the arrangement of the number of black pixels obtained in the first to third embodiments in the (S × T) binary image.
[0072]
FIG. 9 shows a block configuration of an image processing apparatus to which the multi-value image binarization method of the ninth embodiment is applied. In the figure, a scanner unit 1 includes a photoelectric conversion device and an A / D conversion device, reads a document, converts it into multi-valued image data, and outputs it. A binarization unit 2 receives a multi-value from the scanner unit 1. The value image data is converted into binary image data having a density of S times in the main scanning direction and T times in the sub-scanning direction of the reading density of the scanner unit 1, and the printer 3 outputs binary data from the binarizing unit 2. FIG. 1 shows that an image is recorded and output at a density that is S times the scanning density of the scanner unit 1 and T times that of the sub-scanning direction, and the control unit 4 is a microcomputer that controls each of the above parts. This is the same as the image processing apparatus to which the multi-value image binarization method of the first to eighth embodiments shown is applied. The difference is that an edge determination unit 5 is provided.
[0073]
The binarization unit 2 obtains the gradation output value of the target pixel by the simple multi-value conversion processing, multi-value error diffusion processing, or multi-value dither processing of the first to third embodiments described above, and converts from the target pixel. The number of black pixels in the (S × T) binary image is determined by a multi-valued image binarization method in which the number of black pixels in the (S × T) binary image to be obtained is the obtained gradation output value. To do. Further, the binarizing unit 2 determines the arrangement of the number of black pixels determined in accordance with whether or not the target pixel is an edge pixel constituting the edge portion of the image in the (S × T) binary image. To do.
[0074]
The edge determination unit 5 determines whether or not the target pixel is an edge pixel. The edge determination unit 5 determines whether the pixel of interest is an edge pixel, using the same pixel as the pixel that the binarization unit 2 is currently processing as the pixel of interest, and binarizes the determination result. To the conversion unit 2.
[0075]
Specifically, the determination in the edge determination unit 5 as to whether or not the target pixel is an edge pixel is performed by determining each density value of (M × L) peripheral pixels (M and L are natural numbers) around the target pixel and the target pixel. Judgment is based on the absolute value of the difference between the density values.
[0076]
That is, taking the case of M = L = 3 as an example, the density value of the pixel of interest X located at the center of the 3 × 3 pixels shown in FIG. 2 is X, the peripheral pixels A, B, C, D, E, G , F, and H are respectively A, B, C, D, E, G, F, and H, and a predetermined threshold is TE,
| X-B |> TE
｜ X-D ｜ ＞ TE
｜ X-E ｜ ＞ TE
｜ X-G ｜ ＞ TE
When any one of the conditions is satisfied, the edge determination unit 5 determines the target pixel X as an edge pixel. This is because if the target pixel X is an edge pixel, the absolute value of the difference between the pixel values of the neighboring pixels B, D, E, and G that are adjacent pixels and the pixel value of the target pixel is relatively large. If X is not an edge pixel, this is based on the fact that the absolute value of the difference between the pixel values of neighboring pixels B, D, E, and G that are adjacent pixels and the pixel value of the target pixel is relatively small.
[0077]
FIG. 10 shows a processing procedure for converting multi-value image data input from the scanner unit 1 into high-density binary image data in the binarization unit 2 of the image processing apparatus shown in FIG.
[0078]
In FIG. 10, the binarization unit 2 takes each pixel constituting the input multi-valued image data as a target pixel and sequentially captures the target pixel and peripheral pixels (processing 101). Then, the number of black pixels in the binary pixel to be converted from the pixel of interest (S × T) is determined by the multilevel image binarization method of the first to third embodiments described above (processing 102).
[0079]
Then, from the notification of the determination result from the edge determination unit 5, it is determined whether or not the target pixel is an edge pixel (determination 103). When the pixel of interest is an edge pixel (Yes in determination 103), the black pixel position in the binary pixel is determined with priority on the edge (S × T) by the multi-value image binarization method of the fourth embodiment. (Process 104). When the pixel of interest is not an edge pixel (No in determination 103), the (S × T) binary pixel is given priority to the photograph portion according to the predetermined priority by the multi-value image binarization method of the fifth embodiment. The position of the black pixel is determined (process 105).
[0080]
In this way, the binarization unit 2 outputs (S × T) binary pixels in which the number of black pixels included and the arrangement of the black pixels are determined (processing 106).
[0081]
By repeating the above processing as long as there are unprocessed pixels (No determination 107 loop), the binarization unit 2 converts the input multi-valued image into a high-density binary image and outputs it.
[0082]
Even when the input multi-valued image is obtained by reading a document in which characters and photographs are mixed with the scanner unit 1, the clarity of the edges of the pixels constituting the edges of the characters is preserved. On the other hand, for the pixels that are converted into (S × T) binary pixels and constitute the halftone portion of the photograph, the character portion is used to convert the pixels into (S × T) binary pixels while preserving the smoothness of the halftone. The multi-value image can be binarized with high density while satisfying both the image quality of the image and the image quality of the photographic part. Further, the threshold value TE in the edge determination unit 5 is determined according to the number of gradations of the input multi-valued image. In addition, priority is given to the image quality of the character portion or the image quality of the photo portion. The setting can be changed according to whether priority is given.
[0083]
In the embodiment described above, the higher the density value of each pixel constituting the multi-valued image to be converted into the high-density binary image, the blacker the pixel, and the lower the density value, the whiter the pixel. Although the present invention was applied to the case of correspondence, since it is only a convenient correspondence, the opposite correspondence, that is, the lower the density value, the darker the pixel, the higher the density value, the more Even in the case of white correspondence, it goes without saying that the present invention can be similarly applied by rereading the correspondence.
[0084]
Further, in the embodiment described above, when the binarization unit 2 and the edge determination unit 5 refer to the density values of the peripheral pixels of the target pixel, several lines before and after the main scanning line including the target pixel are stored. The binarization unit 2 and the edge determination unit 5 need to include a line buffer memory. However, compared to the case where it is necessary to provide a page memory for binarization processing of a multi-valued image as in the past, it is realized at a much lower cost to have a line buffer memory for several lines. It can be done.
[0085]
【The invention's effect】
According to the first aspect of the present invention, the result of performing simple binarization processing for converting the density value of the target pixel into (S × T + 1) gradations with (S × T) fixed threshold values. The obtained gradation output value includes, as the number of black pixels among (S × T) binary pixels converted from the pixel of interest, the pixel of interest includes black pixels corresponding to the number of gradation output values ( Since the multi-valued image is converted into a high-density binary image by converting into (S × T) binary pixels, a page memory for converting the pixel of interest into (S × T) binary pixels is unnecessary. The multi-valued image can be converted into a high-density binary image by simple processing. In addition, since the number of black pixels to be converted is determined by simple multi-value processing, a high-density binary image output suitable for a multi-value image obtained by reading a character document can be obtained. Also, It is determined whether or not the pixel of interest is an edge pixel from the distribution of density values of (M × N) peripheral pixels (M and N are natural numbers) of the pixel of interest, and when the pixel of interest is an edge pixel In order to determine the arrangement position of the black pixels included in the (S × T) binary pixels converted from the target pixel according to the distribution state of the density values of the (M × N) peripheral pixels. The edges in the multi-valued image can be converted into a fine high-density binary image, and a high-density binary image output that clearly reproduces the outline of characters in the multi-valued image can be obtained. On the other hand, when the target pixel is not an edge pixel, the arrangement position of the black pixel included in the (S × T) binary pixels converted from the target pixel is set according to a predetermined priority. Therefore, the black pixels included in the (S × T) binary pixels converted from the target pixel are regularly arranged, and the multi-level image obtained by reading the photograph has a regular gradation. And can be converted into a smooth and high-density binary image. But Tsu Thus, even a multi-valued image obtained by reading a document in which characters and photographs are mixed can be converted into an optimal high-density binary image.
[0086]
According to the second aspect of the present invention, a gradation output value obtained as a result of performing (S × T + 1) gradation multilevel error diffusion processing on the density value of the target pixel is converted from the target pixel. In order to convert the pixel of interest into (S × T) binary pixels including only the gradation output value black pixels as the number of black pixels of (S × T) binary pixels The multi-valued image can be converted into a high-density binary image by a process that does not require a page memory to convert the pixel of interest into (S × T) binary pixels. Further, since the number of black pixels to be converted by the multi-value error diffusion process is determined, a high-density binary image output suitable for a multi-value image obtained by reading a photographic document can be obtained. Also, It is determined whether or not the pixel of interest is an edge pixel from the distribution of density values of (M × N) peripheral pixels (M and N are natural numbers) of the pixel of interest, and when the pixel of interest is an edge pixel In order to determine the arrangement position of the black pixels included in the (S × T) binary pixels converted from the target pixel according to the distribution state of the density values of the (M × N) peripheral pixels. The edges in the multi-valued image can be converted into a fine high-density binary image, and a high-density binary image output that clearly reproduces the outline of characters in the multi-valued image can be obtained. On the other hand, when the target pixel is not an edge pixel, the arrangement position of the black pixel included in the (S × T) binary pixels converted from the target pixel is set according to a predetermined priority. Therefore, the black pixels included in the (S × T) binary pixels converted from the target pixel are regularly arranged, and the multi-level image obtained by reading the photograph has a regular gradation. And can be converted into a smooth and high-density binary image. But Tsu Thus, even a multi-valued image obtained by reading a document in which characters and photographs are mixed can be converted into an optimal high-density binary image.
[0087]
According to the invention of claim 3, the gradation output value obtained as a result of performing the (S × T + 1) gradation multi-value dither processing on the density value of the pixel of interest is converted from the pixel of interest. As the number of black pixels in the (S × T) binary pixels, the target pixel is converted into (S × T) binary pixels including only the gradation output value black pixels. The multi-valued image can be converted into a high-density binary image by a process that does not require a page memory to convert pixels into (S × T) binary pixels. In addition, since the number of black pixels to be converted is determined by multi-value dither processing, a high-density binary image output suitable for a multi-value image obtained by reading a photographic document can be obtained. Also, It is determined whether or not the pixel of interest is an edge pixel from the distribution of density values of (M × N) peripheral pixels (M and N are natural numbers) of the pixel of interest, and when the pixel of interest is an edge pixel In order to determine the arrangement position of the black pixels included in the (S × T) binary pixels converted from the target pixel according to the distribution state of the density values of the (M × N) peripheral pixels. The edges in the multi-valued image can be converted into a fine high-density binary image, and a high-density binary image output that clearly reproduces the outline of characters in the multi-valued image can be obtained. On the other hand, when the target pixel is not an edge pixel, the arrangement position of the black pixel included in the (S × T) binary pixels converted from the target pixel is set according to a predetermined priority. Therefore, the black pixels included in the (S × T) binary pixels converted from the target pixel are regularly arranged, and the multi-level image obtained by reading the photograph has a regular gradation. And can be converted into a smooth and high-density binary image. But Tsu Thus, even a multi-valued image obtained by reading a document in which characters and photographs are mixed can be converted into an optimal high-density binary image.
[Brief description of the drawings]
FIG. 1 is a diagram showing a block configuration of an image processing apparatus to which a multi-value image binarization method according to an embodiment of the present invention is applied.
FIG. 2 is a diagram illustrating an arrangement of a pixel of interest X at the center of a 3 × 3 matrix and eight peripheral pixels A, B, C, D, E, F, G, and H of the pixel of interest X;
FIG. 3 is a diagram illustrating an arrangement of four binary pixels a, b, c, and d in a (2 × 2) binary pixel matrix converted from a pixel of interest X.
FIG. 4 is a diagram illustrating 16 arrangements of black pixels (or white pixels) in a binary pixel of a 2 × 2 matrix.
FIG. 5 is a diagram showing a 4 × 4 variation threshold matrix;
FIG. 6 is a diagram illustrating an example of a 2 × 2 priority matrix;
FIG. 7 is a diagram showing a priority matrix that is symmetrical to the 2 × 2 priority matrix shown in FIG. 6;
FIG. 8 is a diagram illustrating an example of a 4 × 4 priority matrix in which a high-density binary image after conversion becomes a halftone image.
FIG. 9 is a diagram showing a block configuration different from that of FIG. 1 of the image processing apparatus to which the multi-value image binarization method according to the embodiment of the present invention is applied.
10 is a diagram showing a processing procedure for converting a multi-valued image into a high-density binary image performed in the binarization unit of the image processing apparatus shown in FIG. 9;
[Explanation of symbols]
1 Scanner section
2 Binarization part
3 Printer
4 Control unit
5 Edge judgment part

Claims (3)

Each pixel of a multi-value image composed of pixels having multi-value density values is regarded as a pixel of interest, and the pixel of interest is divided by a main scanning direction S (S is a natural number) times and a sub-scanning direction T (T is a natural number) times. In the multi-value image binarization method for converting the multi-value image into a high-density binary image by converting it into (S × T) binary pixels,
The gradation output value obtained as a result of performing a simple multi-value conversion process for converting the density value of the pixel of interest into (S × T + 1) gradations with (S × T) fixed threshold values is used as the attention output value. As the number of black pixels among (S × T) binary pixels converted from pixels, the pixel of interest is converted into S × T binary pixels including black pixels of only the gradation output value. with, (M × N) peripheral pixels before Symbol target pixel (M, N are natural numbers) from the distribution of the density value of the target pixel is determined whether an edge pixel, the target pixel is an edge pixel In this case, the arrangement positions of the black pixels included in the (S × T) binary pixels converted from the target pixel are set according to the distribution state of the density values of the (M × N) peripheral pixels. On the other hand, if the target pixel is not an edge pixel, the (S × T) 2 converted from the target pixel. Multi-level image binarization how to characterized in that the positions of black pixels contained in the value pixel is determined according to a predetermined priority. Each pixel of a multi-value image composed of pixels having multi-value density values is regarded as a pixel of interest, and the pixel of interest is divided by a main scanning direction S (S is a natural number) times and a sub-scanning direction T (T is a natural number) times. In the multi-value image binarization method for converting the multi-value image into a high-density binary image by converting it into (S × T) binary pixels,
The (S × T) number of binary values converted from the target pixel are converted into gradation output values obtained as a result of performing (S × T + 1) gradation multilevel error diffusion processing on the density value of the target pixel. as the number of black pixels of the pixels, including the black pixels of the pixel of interest by the tone output values pieces together into a (S × T) pieces of binary pixels, before Symbol target pixel (M × N) It is determined whether or not the pixel of interest is an edge pixel from the distribution of density values of peripheral pixels (M and N are natural numbers). If the pixel of interest is an edge pixel, the pixel converted from the pixel of interest is converted. The arrangement position of black pixels included in (S × T) binary pixels is determined according to the distribution state of density values of the (M × N) peripheral pixels, while the target pixel is not an edge pixel Includes an arrangement of black pixels included in the (S × T) binary pixels converted from the target pixel. Position, multi-value image binarization way to and determining according to a predetermined priority. Each pixel of a multi-value image composed of pixels having multi-value density values is regarded as a pixel of interest, and the pixel of interest is divided by a main scanning direction S (S is a natural number) times and a sub-scanning direction T (T is a natural number) times. In the multi-value image binarization method for converting the multi-value image into a high-density binary image by converting it into (S × T) binary pixels,
(S × T) number of binary pixels converted from the target pixel of the gradation output value obtained as a result of performing (S × T + 1) gradation multi-value dither processing on the density value of the target pixel as the number of black pixels of, converts the target pixel (S × T) pieces of binary pixels including black pixels of the tone output values pieces only, before Symbol target pixel (M × N) around It is determined whether or not the pixel of interest is an edge pixel from the distribution of density values of pixels (M and N are natural numbers). If the pixel of interest is an edge pixel, the pixel of interest converted from the pixel of interest ( The arrangement position of the black pixels included in the (S × T) binary pixels is determined according to the density distribution of the (M × N) peripheral pixels, while the target pixel is not an edge pixel. Is an arrangement of black pixels included in the (S × T) binary pixels converted from the target pixel Multi-level image binarization how to characterized in that the location is determined according to a predetermined priority.

Images