JPH09214783A - Picture processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain binary picture data with a fevorable granual sense by reducing the superimposition of plural colors with respect to a picture where smoothness is required while holding clearness in the edge and the contour part of the picture. SOLUTION: Binary picture data concerning the respective colors are generated from respective multi-level picture signals of the plural colors and stored in a storage means 33 together with the respective multi-level picture signals of the plural colors. A binary data re-distribution processing part 34 is provided, where the local feature quantity of the picture is calculated through the use of the multi-level picture signals concerning the plural colors stored in the storage means 33, the distribution of the data values of the plural colors in a peripheral area adding a picture element to be considered is detected through the use of binary picture data stored in the store means 33, it is discriminated whether or not data of the picture element to be considered is distributed to the peripheral picture elements based on the detection result and the feature quantity of the picture in the peripheral area adding the picture element to be considered and data of the picture element to be considered is distributed to the peripheral picture element in accordance with the discrimination result.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing device, and more particularly to a device for generating binary image data to be supplied to an image output device such as a color binary printer.

[0002]

2. Description of the Related Art For example, most of ink jet type and thermal transfer type color printers are so-called binary printers, and when outputting an image, each pixel position to be printed is colored or not colored. Make sure to select either

In this type of binary printer, an error diffusion method is representative of a halftone generating method for generating a binary image from a multi-tone image and artificially expressing the multi-tone image. It has been known.

This error diffusion method is described in, for example, "Journal of the Institute of Image Electronics Engineers, No. 20, No. 5 (1991), pages 476 to 482.
As described in “Page”, the selection of whether to add color to the pixel value (multi-tone image data) of the target pixel of an input image is performed by selecting the error between the input pixel value and its neighboring pixels ( This is performed depending on whether the sum of the weighted average of the difference between the display brightness of the peripheral pixels and the input brightness exceeds a threshold value. That is, the target pixel converts multi-tone pixel data into binary pixel data.

In this way, the multi-gradation pixel data is converted into 2
The error between the input pixel value and the output pixel value caused by the conversion into the value pixel data is obtained as a weighted average of the difference between the already binarized neighboring pixels and the input pixel value. Therefore, the choice of whether to color certain input pixels is chosen to offset the error from the neighborhood cursor.

When this error diffusion method is applied to a color output image, each color for obtaining the color output image, for example, C (cyan), M (magenta), yellow (Y) and black (K) is used. The image is independently applied to each image frame to generate a halftone, and a color image colored according to the generated binary image signal for each color is output.

As described above, in an image in which halftones are independently generated for each color by the error diffusion method and output, what kind of overlap each color of C, M, Y, and K has at a certain position. The output image is not controlled depending on whether or not it is. Therefore, there is a problem of overlapping of colors.

For example, in a light skin color portion of a human face image, cyan and magenta are overlapped with each other to generate a secondary color such as blue and a tertiary color of C, M and Y, that is, process black, which is output. The graininess of the image may be adversely affected.

As a method for avoiding this color overlap,
Although a dot non-color mixing method is known, this method has a problem that the density of the color component that is not output as dots (pixels) is not saved on the screen and the density of the entire image cannot be saved.

As one method for avoiding this problem, Japanese Patent Laid-Open No. 7-107305 discloses that for a two-color image of red and black, when there is an overlap in display dots (pixels), the color with the larger value is used. A method is disclosed in which the color of a pixel is used and the color component having a smaller value that is not output as the color of the pixel is diffused to peripheral pixels in the vicinity of the pixel.

[0011]

However, when the method described in the above publication is used, the generation amount of the secondary color and process black is suppressed to the minimum, but the edge portion and the contour portion of the image are reduced. Also in, since the color is diffused after binarization, the edges of these images,
There was a problem that "blurring" would occur in the contour portion.

In view of the above points, the present invention reduces the overlapping of a plurality of colors for an image that requires smoothness such as skin color while maintaining the sharpness of the edges and contours of the image. An object of the present invention is to provide an image processing device capable of generating binary image data for outputting an image capable of giving a preferable graininess.

[0013]

In order to solve the above problems, the image processing apparatus according to the present invention converts an input color image signal into binary image data by converting the intermediate color image data into binary image data by making reference numerals of the embodiments described later correspond to each other. An image processing device that generates a tone and supplies the image output device, and from each of the multi-tone image signals for each of a plurality of colors for reproducing a color image corresponding to the input color image signal, A binarizing means (32) for generating binary image data for each color, a multi-tone image signal for each of a plurality of colors for reproducing the color image, and a binary image for each color. Storage means for storing data and (33)
And a feature quantity calculating means (342) for calculating a local feature quantity of the image using the multi-gradation image signal stored in the storage means, and binary image data stored in the storage means. By using the detection means (341) for detecting the distribution of data values of at least two or more colors in the peripheral region including the pixel of interest, the detection result of the detection means, and the feature quantity calculation means A discriminating means (34) for discriminating whether or not to distribute the target pixel data to the peripheral pixels based on the feature amount of the image of the peripheral region including the target pixel.
3) and a distribution unit (344) for distributing the target pixel data to peripheral pixels according to the determination result of the determination unit.

[0014]

According to the present invention having the above-mentioned structure, in the binarizing means, from each of the multi-gradation image signals for each of a plurality of colors for reproducing a color image corresponding to the input color image signal, Binary image data for each color is generated. Then, the generated binary image data and the multi-tone image signal are stored in the storage means.

The multi-gradation image signal before binarization for the plurality of colors stored in the storage means is used to calculate the local feature quantity of the image in the feature quantity calculation means. From the calculated feature amount, it is possible to determine whether the image portion including the pixel of interest and its neighboring pixels is the edge or contour portion of the image.

Further, the detection means uses the binary image data stored in the storage means to detect the distribution of data values of a plurality of colors of at least two colors in the peripheral area including the pixel of interest.

In the determination means, when it is detected as a result of the detection by the detection means that there is a color overlap in the target pixel, the target pixel data is distributed to the peripheral pixels from the feature amount from the feature amount calculation means. It is determined whether to do. The distribution unit distributes the pixel data of interest to the peripheral pixels when the determination result of the determination unit indicates that distribution to the peripheral pixels is necessary.

For example, when the discriminating means discriminates from the characteristic amount from the characteristic amount calculating means that the local image in the vicinity of the pixel of interest is not the edge portion but the portion where the reduction of graininess should be prioritized, the attention is paid. Pixel data is distributed to its surrounding pixels. When the local image near the target pixel is an edge portion, the sharpness of the edge of the image is maintained without distributing the target pixel data to the peripheral pixels.

Therefore, according to the present invention, overlapping of a plurality of colors is reduced and a preferable graininess is obtained for an image portion which requires smoothness such as skin color while maintaining the sharpness of the edge of the image. It is possible to obtain binary image data capable of outputting an image giving

[0020]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of an image processing apparatus according to the present invention will be described below with reference to the drawings.

[First Embodiment] FIG. 2 is a block diagram showing an outline of an image output system to which an image processing apparatus according to the present invention is applied. The image output system of this example includes a color image signal generator 1 and a color converter / color corrector 2.
And a halftone generator 3 and a binary color printer 4.

The color image signal generator 1 includes red (R),
A color image signal represented by L ^* , a ^* , and b ^* signals of the three primary color signals of green (G) and blue (B) or the CIE display color system is generated. The color image signal generator 1 can be configured by a computer including an image reading device such as a scanner. The color image signal may be, for example, an image signal obtained as a result of being read by an image reading device, or a color image signal created and edited by so-called computer graphics.

The color converter / color corrector 2 outputs the above-mentioned color image signals to the printer 4 for printing, for example, cyan (C), magenta (M), yellow (Y) and black (K). Is converted into a multi-tone image signal of, for example, 8 bits, which represents the output ratio. In this case, when the color image signal from the color image signal generator 1 is a multi-gradation image signal of each color of cyan, magenta, yellow and black, this color conversion / color corrector 2 is not always necessary. . Since the processing in the color conversion / color corrector 2 is well known and is not related to the image processing apparatus according to the present invention, detailed description of the color conversion / color corrector 2 will be omitted.

The halftone generator 3 obtains C, M for obtaining a binary image output from the printer 4 from the C, M, Y, K multi-tone image signals from the color conversion / color corrector 2. , Y, K binary image data are obtained. The present invention mainly relates to this halftone generator 3.

The binary color printer 4 is composed of, for example, an ink jet printer, and C from the halftone generator 3
An image for each color is generated based on each of the binary image data of M, Y, and K, and a color output image is obtained on, for example, a recording paper as a combination thereof.

In this embodiment, the halftone generator 3 executes a series of processing functions as shown in FIG. 1 to convert the binary image data of C, M, Y and K as described above. obtain. That is, FIG. 1 is a functional block diagram of the halftone generator 3, and in this embodiment, the halftone generator 3 is used.
Includes an image input unit 31, an image binarization processing unit 32, an image information storage unit 33, and a binary data redistribution processing unit 34.

The image binarization processing unit 32 includes an image input unit 31.
The C, M, Y, and K multi-tone image signals input through are subjected to binarization processing independently using an error diffusion method, and the obtained binarized data is used as the image information storage unit 33.
Transfer to As the error diffusion method in this case, the well-known method described above is used. For example, the pixel data of each multi-tone image signal is sequentially added with the error calculated from the neighboring pixels, and then compared with the threshold value. The pixel data (target pixel data) is binarized according to the size.

The processing in the image binarization processing unit 32 is performed in units of lines in this embodiment. In principle, the binarization process can be performed on the entire image at once, or can be performed in units of a predetermined rectangular block. However, the memory capacity required for the image information storage unit 33 is required. This is done in line units in this embodiment from the viewpoint that the number of prints can be reduced and high-speed print processing can be performed.

However, as will be described later, in this embodiment, in the binary data redistribution processing unit 34, binary data redistribution processing is performed on the vertical × horizontal = 3 × 3 = 9 pixels centered on the pixel of interest. Therefore, the binarization processing in units of three lines to be processed is configured to be completed before the processing of the data of that portion is started by the binary data redistribution processing unit 34. .

The image information storage unit 33 includes a binarization processing unit 32.
Each of the C, M, Y, and K binary image data is stored, and the C, M, Y, and K multi-tone image data through the image input unit 31 is stored. As will be described later, the binary data redistribution processing unit 34 causes the binarization processing unit 3 to operate.
Each of the binary image data of C, M, Y, and K from 2 is rewritten as necessary.

The binary data redistribution processing unit 34 is a main part of this embodiment, and the binary image data obtained by the image binarization processing unit 32 can be obtained as a more preferable print output. To be corrected.

In this embodiment, the binary data redistribution processing section 34 performs a processing of subordinately redistributing the cyan image after the image binarization processing to the magenta image. However, redistribution is not performed at the contour or edge portion of the image. That is, whether or not to redistribute cyan pixels (pixels colored with cyan) is determined by the presence / absence of magenta pixels (pixels colored with magenta) at the same position and the image for detecting the contour or edge portion of the image. Judgment based on the feature amount. Specifically, when the cyan pixel and the magenta pixel overlap, if the image portion is a smooth image portion, the cyan pixel is moved to a neighboring pixel position in the vicinity thereof, and redistribution is performed to If the portion is an edge portion, redistribute is not performed even if the cyan pixel and the magenta pixel overlap.

The reason why the cyan image is redistributed subordinately to the magenta image is that magenta has a stronger granular feeling to humans than cyan and yellow. This is to prevent the feeling from getting worse.

In this embodiment, redistributing is not performed for yellow pixels (pixels colored with yellow) and black pixels (pixels colored with black). However, whether to redistribute the yellow pixels is judged based on the magenta pixel and the cyan pixel at the same position,
Redistributing the yellow pixels may be performed. However, since the graininess that humans feel due to the influence of yellow is extremely small compared to magenta and cyan, the effect of the present invention can be achieved without performing the redistribute processing on the yellow pixels as in this embodiment. Is fully obtained.

In this case, the binary data redistribution processing unit 3
4, when the processing function is divided into blocks, as shown in FIG. 1, a binary data duplication detection unit 341, an image feature amount calculation unit 342, a binary data distribution determination unit 343, and a binary data distribution unit 344. , And a random number generator 345.

The binary data duplication determination unit 341 determines whether or not the cyan pixel and the magenta pixel at the pixel position of interest, which is the object of the redistribution processing, overlap each other.
The determination is performed using the binarized cyan image data of 3 and the binarized magenta image data. If there is no overlap, the binary data redistribution processing unit 34 does not perform the distribution process,
The processing described below is not executed.

When the cyan pixel and the magenta pixel are overlapped at the pixel position of interest, the binary data duplication discrimination unit 341 is also arranged near the pixel position of interest.
Check if cyan and magenta pixels are present.

In this embodiment, as will be described later, the range of redistribution is limited to eight pixels R0 to R7 around the target pixel Pz as shown in FIG.
The value data duplication determination unit 341 determines that, for a local area AR consisting of vertical × horizontal = 3 × 3 = 9 pixels, which is equal to the redistribution range and is centered on the target pixel position, a cyan pixel,
Check for the presence of magenta pixels.

Then, in this case, it is determined whether or not cyan pixels and magenta pixels are present, and neighboring pixel information fo in a 3 × 3 matrix format centered on the pixel position of interest.
It is represented by (i, j) (i = 0,1,2; j = 0,1,2). The data value of the cyan pixel at a certain pixel position (x, y) is Cb (x, y), and at the certain pixel position (x, y)
The data value of the magenta pixel in Mb (x, y) is Mb (x, y), and each binary pixel data has a data value of "1" when coloring is performed and a data value of "0" when coloring is not performed. , The neighboring pixel information fo
(I, j) is calculated as fo (i, j) = Cb (x-1 + i, y-1 + j) .OR.Mb (x-1 + i, y-1 + j) (1). Here, * OR * has shown the logical sum.

For example, the cyan binary image data value Cb (x, y) near the target pixel position (xo, yo) is shown in FIG. 4A.
And the pixel position of interest (xo,
magenta binary image data value Mb (x,
When y) is as shown in FIG. 4B, the binary data C in the portion surrounded by the thick line frame in FIGS. 4A and 4B.
Neighboring pixel information fo (i, j) is obtained from b (x, y) and Mb (x, y) as shown in FIG. 4C.

As shown in FIG. 4C, in the nine pixels including the pixel at the target position, the neighboring pixel information fo
Each value of (i, j) is "1" for a pixel position where coloring is performed on either cyan or magenta, and "0" for a pixel position where coloring is not performed. Therefore, the neighboring pixel information fo (i, j) indicates which pixel around the pixel of interest is colored or not colored. This neighboring pixel information fo
The information of (i, j) is supplied to the binary data distribution determination unit 343.

The image feature amount calculation section 342 uses the cyan multi-tone image signal before binarization stored in the image information storage section 33 to determine whether the target pixel is in a smooth region or an edge. Whether it is in the region is digitized by the feature amount obtained by using the edge extraction operator.

In this embodiment, since the redistribution range is the area AR shown in FIG. 3 described above, the feature amount is calculated for this image area AR.

Although many operators are known as edge extraction operators, in this embodiment, a Sobel (Sobe) as shown in FIGS. 5A and 5B is used.
l) An operator calculates an edge gradient (hereinafter, referred to as an edge amount) as a feature amount. Here, the operator in the x direction (horizontal direction) in FIG. 5A is ex (i, j) in the 3 × 3 matrix format, and the operator in y direction (vertical direction) in FIG. 5B is the ey in the 3 × 3 matrix format. If (i, j), the edge amount ep is the cyan pixel C (x,
y), ep = sqrt {(ΣΣ (ex (i, j) × C (x−1 + i, y−1 + j)) ² + (ΣΣ (ey (i, j) × C (x−1 + i, y-1 + j)) ² } (2) is calculated as sq.
rt indicates a square root operation.

Further, the image feature amount calculation unit 342 sets the value of the edge amount ep obtained as described above to Rep = fe (ep) using the normalizing function fe of the curve characteristic as shown in FIG. A value Rep is obtained by normalizing the edge amount ep in the range of 0 to 1. Normalization function f for this normalization
e is obtained as an appropriate value in advance using a test image so that the edge amount can be associated with the appearance preference. This normalized value Rep is 2
It is supplied to the value data distribution determination unit 343.

The processing in the image feature quantity calculation unit 342 is
Specific examples will be further described. FIG. 4A described above
Of the target pixel position (x
The multi-tone image data of cyan near (o, yo) is shown in FIG. 7A.
7B, the edge amount in the x direction can be calculated by using the Sobel operator of FIG. 5A as shown in FIG. 7B: ΣΣ (ex (i, j) × C (x− 1 + i, y-1 + j) = (-1) * 60 + 0 * 40 + 1 * 23 + (-2) * 6
3 + 0x44 + 2x26 + (-1) x67 + 0x47 +
1 × 28 = −150.

Further, as shown in FIG. 7C, the edge amount in the y direction is calculated by using the Sobel operator shown in FIG. 5B. ΣΣ (ey (i, j) × C (x−1 + i, y−1 + j) = 1 x 60 + 2 x 40 + 1 x 23 + 0 x 63 + 0 x 44
+ 0x26 + (-1) x67 + (-2) x47 + (-
1) x28 = -26.

Therefore, the total edge amount ep is ep = 152.237 from the equation (2).

Then, when the edge amount ep is normalized by the normalizing function fe shown in FIG. 6 as shown in FIG. 8, Rep = 0.85 is obtained.

In the binary data distribution discriminating unit 343, the neighboring pixel information fo (i, from the binary data duplication discriminating unit 341 is acquired.
Based on j) and the normalized edge amount Rep from the image feature amount calculation unit 342, it is determined whether the cyan pixel is redistributed or is output as it is. here,
Redistribution of the cyan pixel means movement of the pixel position of the cyan pixel overlapping with the magenta pixel.

In the binary data distribution discriminating unit 343, with respect to each of the nine pixel positions in the area AR centering on the target pixel, a weighting coefficient which means the probability that the target pixel is moved to the pixel position. p (i, j) (i = 0,
1, 2, j = 0, 1, 2) is obtained as follows.
However, in the following description, for convenience of calculation, n × n = 3 ×
The coefficient p (i, j) in the matrix format of 3 is represented by a one-dimensional array, and the coefficient p (k) (k = i + n) as shown in FIG. 9A.
× j; n = 3 and k = 0, 1, 2, ..., N ² −1)
Shall be used.

In the binary data distribution discriminating unit 343, when the weighting coefficient p (k) is the target pixel position of k = (n ² -1) / 2, p (k) = Rep k ≠ (n ² − At the pixel position of interest 1) / 2, p (k) =! fo (i, j) × (1-Rep) ÷ (n × n−Σfo (i, j)) (3) here,! fo (i, j) is a negative value of the neighboring pixel information fo (i, j), that is, a value obtained by inverting each value “1” and “0” of the neighboring pixel information fo (i, j). Means

In the case of the above example, the neighboring pixel information fo
(I, j) is as shown in FIG. 4C, and Rep
= 0.85, the weighting factor p (k) for each pixel position in the area AR shown in FIG. 9A is, as shown in FIG. 9B, p (0) = p (2) = p (3) = P (6) = p (8) = 0 p (1) = p (5) = p (7) = 0.05 p (4) = 0.85.

As described above, the value of each p (k) is the probability that the cyan pixel at the target pixel position is moved to that pixel position. In this case, the neighboring pixel information fo shown in FIG. 9C
Depending on the value of (i, j), the weighting factor p is applied to the pixel position where the cyan pixel or the magenta pixel already exists.
(K) becomes 0, and redistribution is not performed at the pixel position.

In the above example, the value of the weighting factor p (4) at the position of the pixel of interest is 0.85, and the weighting factor of the peripheral pixels that can be redistributed is 0.05. This indicates that it is better not to move the cyan pixel at the target pixel position because the edge amount is large.

Conversely, the smaller the edge amount ep or Rep, the higher the probability of moving the cyan pixel at the target pixel position to the peripheral pixel position. That is, in a smooth image portion that is not the edge portion or the contour portion of the image, the cyan pixel overlapping the magenta pixel is redistributed to the peripheral pixel positions, and the overlapping is suppressed to the minimum.

Weighting coefficient p (k) obtained as described above
Is provided to the binary data distribution unit 344. The binary data distribution unit 344 redistributes cyan pixels according to the weighting coefficient p (k) as follows.

That is, first, the coefficient p (k) is converted into a cumulative probability density pa (k) as shown in FIG. 10A, which is represented in the form of a one-dimensional array for convenience of calculation. However, the value is set to 0 for the pixel position where the cyan pixel or the magenta pixel exists. That is,

[0059]

[Equation 1] Thus, the cumulative probability density pa (k) is obtained. The binary data duplication determination unit 341 is used for the calculation of the equation (4) of the equation 1.
The neighboring pixel information fo (i, j) from is given to the binary data distribution unit 344.

When applied to the above-mentioned example, each value of pa (k) is pa (0) = 0, pa as shown in FIG. 10B.
(1) = 0.05, pa (2) = 0, pa (3) = 0,
pa (4) = (0.85 + 0.05) = 0.90, pa
(5) = 0.90, pa (6) = 0, pa (7) = 1.
00 and pa (8) = 0.

Next, the binary data distribution unit 344 obtains the random number value R consisting of the values 0 to 1 from the random number generator 345, compares it with the cumulative probability density pa (k), and outputs the kth pixel. It is determined whether the cyan pixel is redistributed to the position, that is, moved to that pixel position. In this case, the position of the redistributed cyan pixel is the position corresponding to the minimum k that satisfies R ≦ pa (k).

Applying to the above example, when the random number value R is, for example, 0.05 or less, the cyan pixel at the target position is moved to the position of pa (1), and the random number value R is 0.05.
If it is ˜0.90, the position of pa (4), that is, the cyan pixel at the target position is not moved. If the random number value R is 0.90 to 0.95, the cyan pixel at the target position is moved to the position of pa (5), and the random number value R is 0.95 to 0.95.
If 1.0, the cyan pixel at the position of interest is moved to the position of pa (7). In the case of this example, since the probability that the random number value R from the random number generator 345 is 0.05 to 0.90 is very high, the cyan pixel does not move.

When the redistribution processing is completed as described above, the binary data distribution unit 344 rewrites the binary cyan image data stored in the image information storage unit 33. Therefore, the rewritten binary image data is output to an image output device such as a printer.

The above-mentioned processing in the binary data redistribution processing section 34 is realized by software. FIG. 11 is a flowchart showing the flow of processing in that case.

That is, it is judged whether or not the processing for all the pixels of the redistribution processing has been completed (step S
1) If not completed, the next pixel (at the start of the process, the first pixel) is set as the target pixel (step S2).

After step S2, the process proceeds to step S3, and using the binary image data stored in the image information storage unit 33, it is determined whether or not the cyan pixel and the magenta pixel overlap at the target pixel position. . If they do not overlap, the process returns to step S1 to set the next pixel. If they overlap, the process proceeds to step S4, and the above-described neighboring pixel information fo (i, j) for nine pixels centered on the pixel of interest is generated.

Next, in step S5, the edge amount ep is used as the feature amount of the image of the area AR of 9 pixels near the target pixel, and the edge amount ep is stored in the image information storage unit 33 as the corresponding cyan multi-tone image data. Calculate using. Further, the edge amount ep is normalized to obtain the value Rep.

Next, in step S6, the weighting factor p (k) is calculated using the neighboring pixel information fo (i, j) and the value Rep.
Is calculated. Next, in step S7, the redistribution operation as described above is performed using the weighting coefficient p (k),
In the next step S8, redistribution processing is executed based on the calculation result, and the cyan binary image data in the image information storage unit 33 is rewritten to the cyan binary image data after the redistribution processing.

Then, the process returns to step S1. When it is determined in step S1 that the processing of all the pixels is completed, the process proceeds to step S9, and the cyan binary image data after the redistribution processing is added to the cyan binary image of the image information storage unit 33. Rewrite the image data. And
This processing routine ends.

As described above, according to this embodiment, the overlap of magenta and cyan is suppressed to a minimum for a smooth image portion, so that it can be output as a preferable image with less graininess in appearance. it can. On the other hand, since the redistribute is not performed on the contour of the image or the edge portion, the contour is not blurred and the color is not blurred at the edge portion. [Second Embodiment] Next, a second embodiment of the image processing apparatus according to the present invention will be described. Also in the second embodiment, the target image output system is
It is the same as that shown in FIG. 2, and the block diagram of the halftone generator 3 which is the main part is also the same as that of FIG.

In the second embodiment, the above-mentioned first embodiment is used.
In comparison with the embodiment of FIG.
Among them, the feature quantity algorithm in the image feature quantity calculation unit 342 and the calculation algorithm of the weighting coefficient of distribution to the matrix in the binary data distribution determination unit 343 are different.

The image feature amount calculation process and the binary data distribution process according to the second embodiment will be described below.

In the second embodiment, the image feature amount calculation unit 342 calculates the edge amount as the image information storage unit 3.
With reference to the cyan multi-gradation image data Cb (x, y) before binarization stored in No. 3, the respective values of the eight pixels around the target pixel and the target pixel in the area AR described above. It is calculated as the difference amount d from the value of.

When this difference amount d is represented by d (i, j) (i, j = 0,1,2) in a 3 × 3 matrix format, this difference amount d (i, j) is d (i, j) i, j) = abs {C (x, y) -C (x-1 + i, y-1 + j)} (5). In the equation (5), abs means absolute value calculation.

Then, in the image feature amount calculation unit 342, the difference amount d (i,
In order to make a correspondence between j) and the appearance preference, FIG.
Normalization is performed using the normalization function fd of the characteristic curve as shown in, and the normalized difference amount is defined as Rd (i, j). As can be seen from FIG. 12, the normalized function fd in this case is the normalized difference amount Rd (i, j) when the difference amount d is small.
Is "1", and when the difference amount d becomes larger than a predetermined value, the value of the normalized difference amount Rd (i, j) is rapidly reduced.

Difference amount Rd obtained by this calculation
(I, j) becomes smaller as the pixel position has a larger difference in pixel value from the cyan pixel to be redistributed, and the probability of redistributing becomes lower. In other words, at the contours and edges of the image,
It is possible to prevent redistributing pixels into the area beyond the edge.

In the binary data distribution discriminating unit 343,
From the difference amount Rd (i, j) calculated by the image feature amount calculation unit as described above and the neighboring pixel information fo (i, j) from the binary data duplication detection unit 341, Similar to the embodiment, the weighting coefficient p (k) representing the probability of redistributing to neighboring pixels is calculated. In this case, the weighting factor p
(K) is p (k) =! fo (i, j) × Rd (i, j) (6) However, k = i + n × j and n = 3. Is calculated as

Further, the weighting factor p (4) of the pixel position of interest
P (4) = 1−Σp (k) (where Σp (k) is p for all k
(Meaning the sum of (k)), and if p (4) is a negative value, p (4) = 0.

Further, p (k) = p (k) / Σp (k) is set and normalized so that the total becomes 1. The normalized weighting factor p (k) = p (i + n ×) thus obtained
j) is, as described above, the position (i,
j) is the probability that the cyan pixel is moved. In the case of this embodiment, the smaller the difference value between the target pixel and the neighboring pixel, the higher the probability that the target cyan pixel will be redistributed to the neighboring pixel. Then, in the same manner as described above, the redistribution is not performed to the pixel position where the magenta pixel or the cyan pixel already exists.

The binary data distribution unit 344 redistributes cyan pixels according to the normalized weighting coefficient p (k). That is, as in the case of the first embodiment, the binary data distribution unit 344 acquires the random number value R from the random number generator 345, and the cyan corresponding to the minimum k that satisfies R <p (k). Redistribute pixels. Then, the cyan binary image data in the image information storage unit 33 is rewritten with the cyan binary image data after the redistribution.

A concrete example of the second embodiment will be described below. That is, in the image feature amount calculation unit 342, for example, the cyan multi-gradation image data of the image information storage unit 33 is as shown in FIG. 13A, and the area around the pixel position of interest (xo, yo) is the center. If AR is the area indicated by the thick line in FIG. 13A,
From the equation (5), the difference value d (i,
j) is calculated.

Then, the image feature amount calculation unit 342 uses the difference value d (i, j) as a normalized difference value as shown in FIG. 13D by using the normalization function fd of the characteristic curve shown in FIG. 13C. Convert to Rd (i, j).

In the binary data distribution discriminating unit 343, as shown in FIG.
The weighted coefficient p (k) represented as corresponding to the pixel position as shown in A is the normalized difference value Rd calculated by the image feature amount calculation unit 342 shown in FIG. 15B.
(I, j) and the binary data duplication determination unit 3 shown in FIG. 15C.
It is calculated using the neighboring pixel information fo (i, j) from 41. FIG. 15D shows each value of the weighting coefficient p (k) of this calculation result. Then, the weighting coefficient p (k) is further normalized so that the sum becomes 1, and the coefficient p (k) as shown in FIG. 15E is obtained.

The binary data distribution unit 344 is shown in FIG.
Random number generator 3 according to the normalized coefficient p (k) of
The cyan pixel redistribution processing is executed using the random number value R from 45, and the cyan binary image data of the image information storage unit 33 is rewritten with the cyan binary image data of the executed result.

By performing the redistribution processing as described above, also in the second embodiment, the overlap of magenta and cyan can be minimized for smooth images, as in the first embodiment. Therefore, it is possible to output binary image data capable of outputting a preferable image with a less grainy appearance. On the other hand, when the difference between the pixel of interest and the neighboring pixel is large, the processing is performed so that the pixel is not redistributed, so that the contour is not blurred and the color is not blurred at the edge portion.

[Modification] Since the image processing according to the present invention can be performed without depending on the conventional error diffusion processing method, the present invention performs processing by any error diffusion processing that can be used for at least a single color. The present invention can be applied to multi-color or full-color binary image data that has been created.

In the above-described embodiment, the output image is composed of cyan, magenta, yellow and black. However, the present invention is an image composed of cyan, magenta and yellow without black, and of two colors. It is also applicable to the case of a multi-color image. It can also be applied to red, green, and blue data for displaying an output image on a monitor display.

[0088]

As described above, according to the present invention, the overlap of binary data of a plurality of colors is controlled on the basis of the feature amount of the original image, so that the smooth image portion has a visually granular appearance. It is possible to output a preferable image with less blurring, and to output a clear image in which blurring and color bleeding are suppressed to a minimum for the contour portion and the edge portion.

[Brief description of drawings]

FIG. 1 is a block diagram of a main part of an embodiment of an image processing apparatus according to the present invention.

FIG. 2 is a diagram showing a system configuration of an embodiment of an image processing apparatus according to the present invention.

FIG. 3 is a diagram showing an example of a range for performing redistribution processing in the embodiment of the image processing apparatus according to the present invention.

FIG. 4 is a diagram showing a specific example of a process for obtaining information on overlapping of color pixels and peripheral pixels thereof in the embodiment of the image processing apparatus according to the present invention.

FIG. 5 is a diagram for explaining the calculation processing of the image feature amount in the embodiment of the image processing apparatus according to the present invention.

FIG. 6 is a diagram for explaining a process of normalizing an image feature amount in the embodiment of the image processing device according to the present invention.

FIG. 7 is a diagram for explaining a specific example of image feature amount calculation processing in the embodiment of the image processing apparatus according to the present invention.

FIG. 8 is a diagram for explaining a specific example of a process for normalizing an image feature amount in the embodiment of the image processing device according to the present invention.

FIG. 9 is a diagram for explaining a specific example of calculation of an index as to whether or not redistribution is performed in the embodiment of the image processing apparatus according to the present invention.

FIG. 10 is a diagram for explaining a specific example of an index as to whether or not redistribution is performed in the embodiment of the image processing apparatus according to the present invention.

FIG. 11 is a flowchart showing an example of a processing flow of a main part in the embodiment of the image processing apparatus according to the present invention.

FIG. 12 is a diagram for explaining a process of normalizing an image feature amount in another embodiment of the image processing device according to the present invention.

FIG. 13 is a diagram for explaining a process of calculating an image feature amount in another embodiment of the image processing device according to the present invention.

FIG. 14 is a diagram for explaining a specific example of calculation of an index as to whether or not to redistribute pixels in another embodiment of the image processing apparatus according to the present invention.

[Explanation of symbols]

 1 Color Image Signal Generator 2 Color Converter / Color Corrector 3 Halftone Generator 4 Binary Printer 32 Image Binarization Processor 33 Image Information Storage 34 Binary Data Redistribution 341 Binary Data Duplication Discrimination 342 Image feature amount calculation unit 343 Binary data distribution determination unit 344 Binary data distribution unit 345 Random number generator

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Toru Tanaka Toru Tanaka 430 Nakai-cho, Ashigaragami-gun, Kanagawa Green Tech Nakai Fuji Xerox Co., Ltd. Xerox Co., Ltd.

Claims (4)

[Claims] 1. An image processing apparatus for converting an input color image signal into binary image data to generate a halftone and supplying the halftone to an image output apparatus, wherein a color image corresponding to the input color image signal is reproduced. Binarizing means for respectively generating binary image data for each color from each of the multi-gradation image signals for each of the plurality of colors, and each of the plurality of colors for reproducing the color image. Storage means for storing the multi-gradation image signal and the binary image data for each color, and the multi-gradation image signal stored in the storage means to determine a local image feature amount. Using the characteristic amount calculation means for calculating and the binary image data stored in the storage means, a detection method for detecting a distribution of data values of a plurality of colors of at least two colors or more in the peripheral region including the target pixel. Whether the target pixel data is distributed to the peripheral pixels based on a step, the detection result of the detection unit, and the feature amount of the image of the peripheral region including the target pixel calculated by the feature amount calculation unit. An image processing apparatus, comprising: a determining unit that determines whether the target pixel data is distributed to peripheral pixels according to a determination result of the determining unit. 2. The distributing means, when distributing the target pixel data to the peripheral pixels according to the discrimination result of the discriminating means, distributes only to the peripheral pixels having no overlap of the two or more colors. The image processing apparatus according to claim 1, wherein the image processing apparatus is configured to do so. 3. The feature quantity calculating means calculates the magnitude of an edge gradient as a feature quantity for the multi-gradation image signal, and the judging means distributes the target pixel data to peripheral pixels. The image processing apparatus according to claim 1, wherein whether or not the weight is represented is represented as a weighting coefficient. 4. The feature amount calculating means calculates a difference value between the pixel data of interest and its neighboring pixel data as a feature amount of the multi-gradation image signal, and the discriminating means calculates the difference value. 2. A weighting coefficient indicating whether or not to distribute pixel data to peripheral pixels.
An image processing apparatus according to claim 1.