JPH01227576A - Image processing device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field] The present invention relates to an image processing apparatus having an output device capable of multi-value output, such as a digital copying machine or a printer.

[Prior art]

When reproducing gray scale images (halftone images) in digital copying machines and the like, systematic dithering is commonly used.

When the original image source is a halftone dot image expressed by area gradation, interference occurs between the halftone dot pattern and the dither pattern for dither processing, resulting in moiré. As a method for preventing the occurrence of moire, for example, as described in Japanese Patent Application Laid-Open No. 120025/1980,
There is a method in which a halftone dot image is smoothed, a periodic pattern of halftone dots is removed, and then dither processing is performed. However, by performing the above-mentioned smoothing process, the sharpness (resolution) of the image decreases, and especially if there are letters in the halftone image, the letters become blurred and become unreadable ( It has the disadvantage of becoming.

In addition, recent advances in printers have made it possible to modulate at several levels, not only in binary (white or black) but also in 3 and 4 values.When using such a printer, Gradation can be reproduced using the multilevel dither method.

In the multilevel dither method, various dithering processes are performed depending on the use of intermediate level dots.

FIG. 19 is an explanatory diagram of an example of a dark value matrix (dither matrix) pattern for dither processing. Note that the figure shows a three-value dithering method to simplify the explanation.

The method using the dark value matrix shown in FIG.
The next step is to fill in the second dark value.

On the other hand, in FIG. 6B, the first and second threshold matrices are alternately filled with low-level darkness values. Therefore, pixel (-) in the same figure can more faithfully reproduce the level of each input pixel (pixel), so it has better resolution, but it has poorer output stability (adjacent pixels are given the same level of exposure, and all pixels (4×4=16) and then the next exposure level is assigned sequentially, that is, when the difference in exposure brightness between adjacent pixels is small (
(Dispersion value) Poor stability) Since the appearance rate of intermediate level dots is high, the gradation reproducibility of the matrix as a whole is poor.

On the other hand, in the same figure (a), the drum is at an intermediate level.

Since the appearance rate of glyphs is kept low, smooth gradation characteristics can be obtained.

As described above, with the systematic dither method, it is difficult to achieve both gradation and resolution, so methods have been proposed that adaptively change the processing method depending on the type of image.

In other words, for photographic images where gradation is important (density gradation images), six turns with good gradation characteristics as shown in (a) in Figure 19 are used, and when the resolution performance is high for characters, line drawings, etc. For images that are important, a pattern with excellent resolution as shown in the figure (bl), or binarization or multi-value conversion using a fixed darkness value is used.

Among images, for halftone dot images, gradation is important as in photographic images, so a dither pattern with excellent gradation characteristics is used. However, in a halftone image, minute dots are regularly arranged, and gradation is reproduced by changing the diameter of each dot.

In gradation processing using the systematic dither method, interference occurs between the periodic pattern of the dither and the pattern of the halftone original, resulting in moiré.

In order to avoid the occurrence of moiré, conventionally, when processing a halftone dot document, a smoothing process is first performed to remove the periodic component of the halftone dots, and then gradation processing using a dither method is performed. As a result, images with good gradation reproducibility and no moiré can be obtained, but in halftone images (printed matter, etc.), compared to photographic images, characters are often mixed in the gradation image. There was a problem in that the smoothing process and the use of high gradation patterns made these characters illegible.

〔the purpose〕

SUMMARY OF THE INVENTION An object of the present invention is to solve the problems of the prior art described above, and to provide an image processing apparatus that does not cause moiré in color halftone images and has good character reproducibility.

〔composition〕

As mentioned above, the multilevel dither method includes the methods explained in FIGS. 19(a) and 19(b), depending on how dots at halftone levels are arranged. The method shown in FIG. 3(b) can relatively faithfully reproduce the input level for each pixel in steps of the output multi-value number, and the method shown in FIG. Since the values are highly biased, it is difficult to faithfully reproduce the input level on a pixel-by-pixel basis.

Moiré does not occur if the input data is faithfully reproduced pixel by pixel, but it appears particularly prominently when using a periodic pattern in which the threshold values for each pixel are highly biased, as shown in Figure 19(a). . Also, in the case of FIG. 19(b), since a regular pattern is used, moiré does not occur.
Since the input data is reproduced relatively faithfully even at the pixel level, the power of moiré can be kept low. This moire reduction effect becomes more pronounced as the number of multi-values increases (at about 8 values, moire is suppressed to a level where it is hardly visually noticeable. Also, halftone images are essentially binary images. Therefore, even if a pattern that uses many intermediate levels, such as the one shown in FIG. As mentioned above, the unevenness (instability) of characters can be suppressed, and the reproducibility of characters is good.

Therefore, for a halftone image, a higher quality output can be obtained by using the pattern shown in FIG. 19(a) than by using the pattern shown in FIG.

In the present invention, images are classified into three types: photo/halftone dot/text, or two types: photo/halftone dot/text, and for photo images,
A dither pattern emphasizing gradation as shown in (a) in Fig. 19 may be used, and a pattern shown in ('b) in Fig. 19 may be used for halftone images, or in order to further improve the quality. To,
Edge enhancement, binarization or multi-value conversion using a fixed threshold value may also be performed.

Furthermore, the dither pattern is not limited to the dot-distributed type illustrated in FIGS. The dither patterns shown in ) and (b) are extremes in terms of how dots are used (arranged) at intermediate levels, and depending on the performance of the printer, it may be possible to improve gradation and resolution by using intermediate patterns. is possible.

The threshold layout is as shown in FIG. 19(C).
If the dots are concentrated, more stable gradation characteristics can be obtained.

Here, a case will be explained in which the dark value matrix shown in (a) and (bl) of FIG. 19 is expanded to four or more values.

In the multilevel dither method, N dark value matrices are usually used to convert into (N+1) values. This darkness value matrix is numbered from 1 to N.

Also, when the number of elements in the matrix is M, each element has 1
Numbered from ~M. If a square of size mxn is used, M=mX. To be more precise, M is the number of pixels constituting the unit of gradation expression. At this time, the threshold of the first element of the jth threshold matrix is set to thr
(t, j), then thr (t, j) is, In the matrix pattern of the type shown in FIG. 19(a), thr (i, j)=iXN+j-2− )
So, in the matrix pattern of the type shown in Figure 19 (bl), t h r (i, j) = Mx (j -1)
+ i---(21)

(1), (In Equation 21, by changing the correspondence between the element number of the matrix and the position in the matrix, a dot concentrated type or dot dispersed type pattern can be obtained.

Here, a case is described in which the input data is a linear density signal quantized to M levels.

Correction is required if the input data is not linear in density or if the output level is not linear in density.

In particular, when the number of expressible gradations (NXM+1) is greater than the number of levels of input data, the gradation characteristics can be set so that the output density value is linear with respect to the input density value.

Embodiments of the present invention will be described below with reference to the drawings.

First, before entering into a description of processing of multicolor originals, image processing of monochrome (white, black) originals will be described.

FIG. 3 is a system block diagram showing an example of an image processing device for image signals from a monochrome original, which is the premise of the present invention, and (a) shows whether an image is processed into a photo area or a halftone/text area. This is a specific example configuration in which the processing is switched based on the classification.
1 is an input system, 2 is a photo section processing section, 3 is a halftone/character section processing section, 4 is an area determination section, 5 is a selector, and 6 is an output system.

In addition, FIG. 3B shows a specific example in which the image is classified into three types: a photo area, a halftone area, and a character area, and the processing is switched. 3-1 is a processing unit for the halftone area; -2 is a character part processing unit, and 5-1 is a selector.

In the same figure (C), edges are emphasized in the text part,
A tool configured so that edge enhancement processing is not performed in halftone areas.
On the body side, 7 is an edge enhancement processing section, and 8 is a selector.

Note that in (b) and (C), the same symbols as in (a) correspond to the same parts.

In each of the specific examples shown in FIG. 3 above, the processing results by the plurality of image processing units are selected according to the judgment results of the area judgment unit, and as shown in (C), the area judgment unit 4 In addition to selecting the processed results according to
Depending on the area determination signal, the dither threshold data of the photo area processing section and the halftone/character section processing section may be changed, or the calculation of the edge enhancement filter of the edge enhancement processing section may be changed. .

FIG. 4 is a block diagram showing an example of the configuration of the dither processing circuit in the photo section processing circuit and the halftone dot/character section processing section in FIG. 10 is an X-ary counter that receives the pixel synchronization signal; 1
1 is a ROM table containing image data and counters 9 and 10.
By storing the processing result in the ROM address addressed by the address of the threshold matrix whose address is the output of , multi-value dither processing is performed in a table reference manner. Note that here the dither matrix size is y
A case where a rectangle of xx is shown is shown.

The configuration and operation of the image processing apparatus described above will be described in detail below.

First, regarding the area determining section 4, the photograph/halftone/character separation algorithm will be explained.

(1) Judgment based on colored pixel density This judgment is made as follows: (1-1) When binarized at a low level (background level + α), pixels exceeding the background level (= colored pixels)
can be extracted.

(1-2) In general, it is considered that most areas of a density-graded photographic original are composed of pixels at or above the background level.

(Since the 1-3Nl 1-point character image is a black and white binary image, it is thought that the same amount of background-level pixels and high-density colored pixels coexist.

(1-4) Calculating the density of the colored pixels extracted in (1-1) above, based on the assumptions (1-2) and (1-3), the density is almost at the highest level in photographic images, and the density is almost the highest level in photographic images. Point/character images are considered to have medium density.

Therefore, a certain threshold value is set, and when it is equal to or greater than the threshold value, it is determined to be a photographic image, and when it is smaller than that threshold value, it is determined to be a halftone dot/character image.

FIG. 5 is a block diagram for a determination algorithm based on colored pixel density, in which 10 is an MTF correction circuit, 30 is a comparator, 40 is a colored pixel density filter, and 50 is a comparator. This algorithm is a separation method that uses the fact that the density of colored pixels is low at the black-and-white border (edge) of characters, etc., so if the edges are heavily rounded, misjudgments are likely to occur.

Therefore, before the comparator 30 extracts colored pixels, it is preferable to perform MTF correction on the input system.

An MTF correction circuit 20 is provided for this correction.

The comparator 30 has input image data at the background level (thrl).
When the value exceeds , the colored pixel is extracted as a colored pixel.

. The colored pixel density filter 40 may have a size of approximately 4×4 to 8×8.

As explained in (1-2) above, in the photographic area, almost all pixels in the scanning window are colored pixels, so the darkness value (thr2) for area determination by the comparator 50 is Of the number N of reference pixels, the number may be set to about N to N-2. Furthermore, if the scanning window size is, for example, 7×7, this reference pixel is not all 49 pixels, but
The configuration can be simplified by using some pixels as shown below.

FIG. 6 is an explanatory diagram of an example of pixel arrangement of reference pixels, (a
) indicates all 7×7 pixels, crest) indicates a cross shape, (C) indicates an X-shape, (d) indicates a rice layer, (e) indicates a mouth shape, and (f) indicates a concave pixel arrangement.

As explained above with reference to FIG. 1, the above-mentioned determination based on colored pixel density requires correction of MTF deterioration in the input system, particularly in the high frequency region, for image data read by a scanner. Therefore, the MTF correction circuit 20 can be a digital filter with high frequency emphasis characteristics. High frequency enhancement is generally performed by subtracting a Laplacian (second-order differential) from the original image.

FIG. 7 is an explanatory diagram of the two-dimensional high-frequency emphasis filter used for MTF correction, where (a), (bl, and (C) are 3×3
Size filters (d) and (e) are 5×5 size filters. The size of the filter is 5x3 rather than 3x3.
By using a larger size such as X3, 5X5, etc., more appropriate correction can be performed.

The coefficients of the filter are actually determined according to the MTF characteristics of the input system. Also, the MT of the actual input system
The coefficients in the main scanning direction and the sub-scanning direction may be configured to be different depending on the F characteristic.

Note that, as shown in FIG. 4B or FIG. 3C, if only pixels in four directions are used, the hardware configuration can be simplified.

FIG. 8 is a circuit block diagram showing an example of a 3×3 filter.
Line buffer 20-1.20=2 and 9 latches 20-3.20-4.20-5.20-6.20-
7.20-8.20-9゜20-10.20-11, and the adder 20-
12.20-13.20-14.20-15.20-1
6.20-17.20-19.20-20 and multipliers 20-18 are used to perform the calculation as shown in FIG. It should be noted that by using a ROM instead of the adder 1 multiplier, it is also possible to provide a configuration in which matrix operations of arbitrary coefficients can be easily performed.

FIG. 9 is a circuit block diagram showing an example of a colored pixel/element density calculation circuit as a colored pixel density filter. Here, an example is shown in which the number of colored pixels within a 5×5 scanning window is counted. In the figure, 1-bit data that has been converted into a low-level binary value is input to a shift register 40-1. The shift register 40-1 outputs binary data for 5 pixels (5 bits in total) in the main scanning direction. Next stage counter 4
In 0-2, the number of colored pixels per 5 pixels in the main scanning direction is counted. That is, the number of "1"s in 5 bits is counted.

The counter 40-2 uses a ROM and uses a table reference method to store values from 0 to 5 corresponding to the number of "1" in the 5 bits at the address addressed by the 5 bit data. This can be achieved by

The pixel density data in the main scanning direction is stored in the line buffer 40-
3.40-4.40-5.40-6 and latch 40-7.
40-8.40-9.40-10゜5 lines are held by 40-1.1, and the pixel density data for these 5 lines are added to adders 40-12.40-13.40-14 and 40.
By adding -15, the number of colored pixels per 5×5=25 pixels is calculated. This colored pixel density and a predetermined threshold value thr (approximately 25 to 23) are compared by a comparator 40-16, and when the comparison result is larger than the threshold value thr, it is "O" (photograph area), and when it is smaller, it is "1" (halftone dot). Outputs an area determination signal for (character area).

Next, a photograph/halftone/character separation algorithm based on edge pixel density will be explained.

(2) Judgment based on edge pixel density: (2-1) Edge pixels are extracted using Laplacian operator, Roberts operator, etc.

(2-2) In a photographic image (tm degree gradation), there are few sudden changes in density, so there are few pixels from which edges are extracted.

(2-3) Since a character image is basically a binary image, many pixels are subjected to edge extraction.

(2-4) Calculating the density of edge pixels, the above (2-4)
2) and (2-3), the density is low in the photo area and high in the character area.

Through the above steps, photographs/halftone dots/characters are separated.

FIG. 10 is a block diagram showing a configuration example for executing the algorithm (2) above, in which 60 is a differential value calculation circuit, 70 is a comparator, 80 is an edge pixel density filter,
90 is a comparator.

In the same figure, the Laplacian (
A common method is to use the magnitude of the gradient (second-order differential) or gradient (first-order differential) value. The result of this differential value calculation is compared with a threshold value thr3 in a comparator 70, and when the calculation result is larger than the dark value, it is determined to be an edge, and when it is smaller, it is determined to be a non-edge.

FIG. 11 is an explanatory diagram of an example of a differential filter, (al
~(d) is a second-order differential filter, and (el~(n) is a first-order differential filter.

In edge extraction, the magnitude of the absolute value of the differential value is a problem, but since the first-order differential filters shown in (e) to (nl) have strong directionality, (e) - (f), (phantom ~(h),
(i) −U), (kl−(1), ((2)−(
n), the 2 of the orthogonal differential values f, , fX of the directionality
It is preferable to use the square root square of the sum of products -2. However, for simplicity, '(l fxl ” l fyl
)/2. max(l rXl, I fyl) may also be used. Alternatively, the maximum value of the gradient in the 8-direction at the pixel position of interest may be used.

FIG. 12 is an explanatory diagram for determining the maximum value of the gradient in the eight directions mentioned above. If the gradation level of each pixel in the 3×3 scanning window is a% i as shown in the figure, then the pixel of interest The concentration gradient (first derivative) at the position (position e) is Ia-e Ilce 1max(
,! -9l b e l, -1d el.

J″2: 1g-el l f-e I, -, Ih-e 1. ” ”
)1'T IIIT.

The determination level thr3 for edge extraction by the comparator 70 in FIG. 10 is a value that should be determined based on the MTF characteristics of the input system and the characteristics of the image to be extracted.

This threshold thr3 may be a fixed value, but may also be a floating darkness value that takes into account the background level of the document.

Then, when the output of the differential value calculation circuit 60 is larger than the threshold value thr3, it is assumed to be an edge, and when it is smaller, it is assumed to be a non-edge.

After passing through the edge pixel density filter 8o at the next stage, a comparator 90 uses a threshold value thr4 to determine the areas of photographic areas and halftone/character areas.

Advantages of using edge pixel density include prevention of omission of extraction of halftone dots and character areas due to the influence of noise, and prevention of erroneous determination in photographic areas. The size of the density filter may be approximately 3×3 to 8×8.

In the block diagram of FIG. 10, the differential value output from the first stage differential value calculation circuit 60 and a predetermined threshold value (thr3
) to determine whether it is an edge pixel. The differential value calculation can be realized by using the same MTF correction circuit as shown in FIG. 8 above. The filter coefficients shown in FIG. 11 may be used at that time.

For edge pixel density calculation and area determination, a circuit similar to the colored pixel density calculation and determination circuit shown in FIG. 9 can be used. At this time, when the edge pixel density is greater than a predetermined darkness value (t h r 4), "1" (text area) is output, and when it is smaller, "0" (picture-photograph area) is output.

Next, an area determination algorithm based on halftone dot detection will be described.

(3) Judgment by @ point detection This judgment is for extracting halftone dots and character areas. (3-1) Slice the input image and divide it into two values.

(3-2) Halftone dots (dots) are detected on the binary image using a pattern matching method.

(3-3) For each unit block, it is determined whether or not it is a halftone dot area based on the presence or absence of halftone dots.

(3-4) Correct the determination based on the determination results of adjacent blocks to reduce false determinations due to noise.

FIG. 13 is a block diagram for an area determination algorithm based on halftone detection, in which 100 is an MTF correction circuit, 11
0 is a comparator, 120 is a halftone detection circuit, 130 is an area determination circuit, and 140 is a determination correction circuit.

A halftone image is basically a binary image, and the dot diameter is modulated according to the level of density to be expressed. In this algorithm, this dot is detected, and the vicinity of the area where the dot is detected is determined to be a halftone image area.

A pattern matching method is used to detect dots.
Halftone pitch.

Since the dot diameter changes depending on the dot area ratio, multiple types of templates are prepared.

FIG. 14 is an explanatory diagram of an example of a template used in the pattern matching method, and each element (pixel) within the scanning window is represented by M 1 ). Note that the figure shows three types of templates of different sizes indicated by ○, ×, and Δ.

The halftone dot detection conditions using the template shown in FIG. 14 are as follows. Regarding pixel M44, next, the image is divided into blocks having a size of approximately 8×8 to 16×16, and the block in which a halftone dot is detected is determined to be a halftone dot area.

FIG. 15 is an explanatory diagram of judgment correction processing performed after area judgment in order to prevent misjudgment due to noise etc. For example, among adjacent blocks 1 to 4 of 8 x 8 pixels, 3 or more blocks are halftone dots. If it is determined to be a region, the remaining block is also corrected to be a halftone dot.

Conversely, if the number of blocks determined to be halftone dots among the four blocks is two or less, it is corrected that all four blocks are not halftone dot areas.

Hereinafter, a circuit configuration for executing the above-mentioned algorithm for region determination based on halftone dot detection will be described.

In the block diagram of FIG. 13, the area determination algorithm based on halftone dot detection detects halftone dots from their shapes, as described above, and the detection performance is as follows:
Since it largely depends on the MTF characteristics of the input system, it is desirable to perform MTF correction.

The image data subjected to MTF correction by the MTF correction circuit 100 is binarized by a comparator 110 at a normal binarization level (approximately 32 in the case of 6-bit data) threshold value thr5,
Front point detection is performed using the template explained in FIG. 12 above. The details of this halftone dot detection circuit 120 will be explained.

FIG. 16 is a configuration diagram showing an example of a halftone dot detection circuit,
Figure (a) shows a circuit for aligning 49 pixel data in a 7x7 scanning window, including a line memory 120-1 for 6 lines and a latch 120-1 for 7x7 = 49 bits.
Consists of 7 to 120-19. 49 pixels worth of binary data M, ~M? , can be taken out of the latch group simultaneously.

Figures 16(b), (C), and (d) are halftone dot detection circuits using pattern matching.
d) are the three types of templates shown in Figure 12.○
,×,Δ.

In the same figure, three types of tenbrai) 0. For X and Δ, the determination result n""a + nm is output.

Judgment output (11nil 1 n4) is "0" when it matches each pattern, and "1" when it does not match.

FIG. 5(e) is a circuit diagram for obtaining the halftone dot detection signal n, which outputs "1" when a halftone dot is detected, and "O" when there is no halftone dot.

FIG. 17 is a configuration diagram showing an example of an area determination circuit,
130-1 is an 8-bit serial-to-parallel converter, 1
30-2 is OR, 130-3 is latch, 130-4 is O
R, 130-5 is line memory, 130-6 is latch,
130-7 is a latch.

In the figure, the presence or absence of halftone dots is detected for each block of 8×8 size, and a halftone dot area signal P is output.

If at least one halftone dot is detected in the 8×8 block, P="1" is output, and if no halftone dot is detected, P="0" is output.

Halftone detection signal n from the halftone detection circuit 120 (FIG. 13)
is input to an 8-bit serial/parallel converter 130-1, and every 8 pixels accumulated in the main scanning direction, 8 bits are output in parallel to the 0R 130-2. 0R130-2 outputs "1" if even one of the 8 bits (8 pixels) is 1" (halftone dot detection). In other words,
The presence or absence of halftone dots within the 8×1 block is determined. This 0R1
The output from 30-2 is latch 130-3.0R130-
4 and is stored in the line memory 130-5. The line memory 130-5 stores area determination results for every 8 pixels of the l-th line. When the second line halftone detection signal is input, the 0R 130-2 outputs an area determination signal for every eight pixels of the second line. At the same time, the line memory 130-5 outputs an area determination signal for every 8 pixels of the first line, and the (:) R130-4 outputs an 8×2 block area determination result. Line memory 130-
5 stores the area determination result of the 8×7 block.

When the halftone dot detection signal of the eighth line is input, the area determination result of the 8X1 block of the eighth line is input to the 0R 130-4 through the latch 130-3. At the same time, the area determination results for the 8x7 block from the 1st to 7th lines are displayed in the latch 1.
It is input to 0R130-4 through 30-6. 0R1
30-4 outputs the area determination result of the 8×8 block. Every time the judgment result P of 8×8 block is obtained,
The determination result is held in latch 130-7. 8×8
When the block determination result P is held in latch 130-7, latch 130-3. The output of the latch 130-6 is cleared, and "0" is written in the line memory 130-5, in preparation for area determination of the block on the 9th to 15th lines.

FIG. 18 is a configuration diagram showing an example of a judgment correction circuit,
140-1, 140-2 are line memories, 140-3~
140-6 is a latch, 140-7 to 140-10 are three-man NAND, and 140-11 is four-man NAND.

In the figure, the area signal P from the area determination circuit is stored in a line memory 140-1 every 8 lines. The area signal P stored in the line memory 140-1 is output for every eight pixels and is held in the latch 140-3. Line memory 140-1 performs the same operation for eight lines. That is, the same signal is output while processing an 8×8 block. When the eighth (eighth line) data is output from this line memory 140-1, the data is stored in line memory 140-2.

At the same time, the next block (
9th line) data is stored.

In this way, the line memory 140-1.140-2 has
Area signals of blocks adjacent to each other in the sub-scanning direction are stored. Line memory 140-1.140
-2 outputs an area determination signal every eight pixels, and latches 140-3. It is held at 140-4. latch 1
40-3. The output of 140-4 is the next stage latch 140-
5.140-6 is latched every 8 pixels. These 4
As shown in Figure 15, each latch has three manual NA
When at least 3 of the 4 adjacent blocks are determined to be a halftone area by the NDs 140-7 to 140-10, the block of interest is determined to be a halftone area, and when there are 2 blocks or less, it is determined to be a non-halftone area. judge. Here, when it is determined that the area is a halftone dot area, "1" is output, and when it is determined that it is a non-halftone area, "0" is output from the four-manpower NAND 140-11.

Three types of area determination methods (11, (2), (3)) have been described above.

Both area determination methods (1) and (2) are based on photographic areas/halftone dots/
Since the character area is determined, either one may be used. However, in (1), since the background part is determined to be a text area, the highlighted part of the photograph is likely to be determined to be a text area, but on the contrary, low-density text is determined to be a text area, which is convenient. Furthermore, in (2), the background portion is determined to be a photographic area, which is convenient for the highlight portion of the photograph, but has the characteristic that low-density (low-contrast) characters are likely to be omitted from extraction.

On the other hand, in the determination method (3), a halftone dot/character area is determined, and a continuous gradation (density gradation) photograph is determined to be a character area. In this way, by using the judgment method (1) or (2) and (3), it is possible to separate the image into three types: photo/halftone dots/text, or into two types: photo/W4 dots/text. can.

As described above, the photographic image (density gradation image) and the halftone dot/text image are separated, and the photographic image area is subjected to centralized dither processing, while the halftone dot/text area is subjected to distributed dither processing. By applying dither processing, photographic images are not smoothed more than necessary, and their sharpness is improved. By performing dither processing, moiré can be suppressed and an output image with good sharpness can be obtained.

By performing distributed multi-value dither processing on the halftone character area, density smoothing can be made unnecessary and good results can be obtained for the characters as well.

Note that halftone dots and characters may be further separated.

Furthermore, in the above, when separating an image into two types: photo/halftone dot/text, it is sufficient to use only one of the determination methods (1) or (2), which simplifies the configuration.

As explained above, by separating an image into a density gradation image (photo) and an area gradation image (halftone image or text) and performing gradation processing using a dither pattern suitable for each image, The sharpness of any image can be improved, and even shaded characters can be reproduced sharply.

Next, a detailed explanation will be given of the present invention in which the above-described image processing technology is applied to color image processing of a multicolor copying machine (so-called full-color copying machine) or the like.

In a so-called full-color copying machine, a document serving as an image source is separated into multiple colors, for example, red (R), green (G), and blue (B), and read. For each color of image data, the area is determined by applying the above-described determination method. Thereafter, a method can be considered in which the optimum gradation process is selected for each area according to the respective determination results for each color.

In the present invention, different dither patterns are used for photographic areas (density gradation) and halftone areas (area gradation), but different dither patterns have different T characteristics. In addition to gradation, color reproducibility is also an important factor.Usually, three colors, yellow (Y), magenta (M), and cyan (C), or four colors including black (BK), are used. Three or four dither patterns are used as IMi, but any pattern that is unspecified for even one color is used as one set, but if even one color is used that is unspecified, the T characteristics and the degree of overlapping of color materials may be affected. change.

This causes problems such as color balance being disrupted and color reproducibility becoming poor.

Therefore, in areas where color reproducibility is important, such as halftone areas and photographic areas, it is desirable to use one set of patterns.

Therefore, in the present invention, a comprehensive determination of the area is performed based on the results of area determination for each color. At this time, consider the following points.

1. An area that is determined to be a halftone dot area even for one color is often a halftone image area for all colors (determination result by the above-mentioned determination method (3)).

2. An area that is determined to be a text area (judgment result according to the determination method (2) above) even with one color is likely to be a text area even if it contains a BK acid component.

3. Color When there are characters on the background, the color component of the background color becomes a photographic area, but among the color components of the characters, the color components other than the color component of the background color become a text area.

Based on the above considerations 1 to 3, the following comprehensive judgment is made.

If even one color is determined to be a halftone dot area, all color components are determined to be a halftone dot area.

2. In areas other than the 8-point area, the judgment results for the text/photo area are followed for each color.

3. When reproducing colors using 4 colors, 1
If it is determined that the area is a text area based on the color, it is determined that the area is a text area regarding the BK acid component.

The above is a comprehensive judgment method when classifying an image into three types of areas: text/halftone dots/photo. In this comprehensive judgment method, the judgment result of the halftone dot area is given top priority, but the reason for this is that it is difficult to extract the characters hidden in the halftone image (characters covered by yl), and This is because if a dither pattern of the type shown in FIG. 19 is used as gradation processing for point areas, characters can be reproduced satisfactorily.

Therefore, the same gradation processing may be performed on the character area and the halftone dot area. In item 2 of the comprehensive judgment method above, text/photo area judgment is performed for each area, but
In this case, for photographic images, as mentioned above, patterns outside the dither pattern settings are mixed in, resulting in
Although there is a possibility that color reproducibility will deteriorate, when characters are mixed, giving priority to character reproducibility will result in better overall image quality, so this decision is made.

Item 3 above takes into consideration the improvement of the reproducibility of black characters on colored backgrounds, but BK, BK,
The component is determined to be a character area, but since there is no BK acid component in the case of color characters, there is no negative shadow in the reproduced image regardless of which gradation process is selected for the BK acid component. In addition, for the text/photo area determination in item 2.3 above, the determination method (2) is suitable because the background is determined to be a picture area, but in order to improve the determination accuracy, determination method (1) is suitable. ) may be used together.

Furthermore, a method for further improving image quality will be explained.

In character images, black characters on a white background are frequently used, and improving the quality of black characters greatly contributes to improving overall image quality.

By the way, when reproducing black letters, especially Y.

When using the four colors M, C, and BK, if 100% UCR (undercolor removal) processing can be performed ideally, reproduction will be achieved using only BK. However, in reality, color components remain due to noise and color balance deviations.

This color component colors the reproduced black characters, reduces sharpness, and deteriorates quality. Therefore, (
Do not output black text (on a white background) without removing color components),
It is preferable to reproduce it only in black. The area of black characters on a white background (hereinafter referred to as black and white characters) can be determined by the following method.

Black and white characters are character areas for all R, G, and B color components, so when all three colors are determined to be character areas,
It is determined that the area is a black and white character area.

At this time, the background needs to be determined as the picture area, so
The determination method (2) is used. Note that, in order to improve the accuracy of determining photographic areas, determination method (1) may be used in combination. In this case, in the determination methods (1) and (2), the character area is set as true ("1") and the picture area is set as false ("0"), and the logical product of the two is used for determination.

FIG. 1 is a block diagram showing an embodiment of image processing in the above-mentioned full-color copying machine, in which 1 is an input system, 1-
1 is a color correction circuit, 1-3 is a halftone area judgment circuit based on judgment method (3), 1-4 is a character area judgment circuit based on judgment m + 2) / (1), 1-5 is a color correction/UCR circuit, 1-6
is a tone processing circuit for halftone dots, 1-7 is a tone processing circuit for characters,
1-8 is a photographic gradation processing circuit, 1-9 is a selector (1)
, 1-1O is a selector (2), 1-11 is an output control section,
6 is an output system.

In the figure, the original is first input to R2O and B by the input system 1.
The image is separated into colors and read. This color separation data R, G,
B is input to the color correction circuit 1-1 and the color correction/UCR circuit 1-5.

The color correction/UCR circuit 1-5 performs color correction of the color material of the output system 6 and performs under color removal (UCR) processing for black generation, and outputs four colors of Y, M, C, and BK. The amount of coloring material is calculated. This output data Y, M, C, BK is 3
Gradation processing circuit for seeds (lii point gradation processing circuit 1-61, character gradation processing circuit 1-7, photographic gradation processing circuit 1-8)
In each circuit, optimal processing is performed for each halftone dot character and photographic image.

On the other hand, the color correction circuit 1'-1 performs color correction of ink used for the original. This is especially for improving the determination accuracy for black and white character areas.

-a, among the Y, M, and C inks used for printed matter, M and C contain other color components. Therefore, especially
Secondary colors such as R, G, and B also absorb R, G, and B light, respectively. As a result, such color characters are R,G.

All of the three color components of B are determined to be text areas, and are therefore determined to be black text areas.

In the embodiment shown in the figure, for black and white characters, Y, M
Since the output of the color components of , C is stopped, the erroneously determined color characters will disappear. In order to avoid this problem, masking processing is performed to correct the spectral characteristics of the coloring material of the printed matter. The outputs from the color correction circuit 1-1 are Y and M.

For each color component of C, in the halftone area determination circuit 1-38 and the character area determination circuit 1-4, the determination method (3) 1 determination method (2) is performed.
) or/and (1), the halftone dot area 9 character area is extracted.

The halftone area determination circuit 1-3 receives the input color component Y.

For M and C, when each is in a halftone area, output YHT
I MHt+cal is set to true (11'').Furthermore, the character area determination circuit 1-4 receives the input color component Y.

For M and C, when each is a character area, output Ycm
□1M0□+ CCllAl is played true (“1”).

And B K eMam = Y C)IAII
Or M CIIARorCCHAR, and these outputs YCIIAII+ MCIAII+CCH^RI
Character gradation processing circuit 1-7 according to BKCHAR
The output of the photographic gradation processing circuit 1-8 is selected by the selector 1-9.

, Next, the halftone area signal = (YHT and Y
CHAI) or (MHt and MC'NAI)
According to or (CHy and CCHAR), the selector (2) selects the output of the halftone dot gradation processing circuit 1-6 or the output of the selector (1). Of these, Y, M, C
Each color component is black character area signal = Yc, Inx an
d MCHARand CCHAR and the black and white character area signal obtained by the logical product of the halftone area signal complement, the output control unit 1-11 controls the output Y, M, C, B.
K is sent to the output system 6. In output system 6, Y, M,
An image is reproduced using four color materials, C and BK.

FIG. 2 is a block diagram showing another embodiment of the present invention in which the gradation processing is switched between the halftone dot/text area and the photographic area, and 1-12 is the halftone/text gradation processing. circuit, 1
-13 is a selector, and the same reference numerals as in FIG. 1 correspond to the same parts.

In the figure, the difference in configuration between this embodiment and the embodiment shown in FIG.
The point is that the logical sum of l1All+CCNA1+BKcn□ and the halftone dot area signal is used. The rest of the movements are the first
This is similar to that explained in the figure.

Each of the above embodiments deals with the case where an image is reproduced in four colors (Y, M, C, BK), but three colors (Y, M, C) are reproduced.
When reproducing using color, it is possible to apply the BK capital letter BK excluding the part related to the acid component.

〔effect〕

As explained above, according to the present invention, halftone dot areas and photographic regions are separated for each color component, and even if only one color component is determined to be a halftone dot area, all color components are divided into halftone dot areas. Since gradation processing can be selected, color reproduction is particularly good for halftone originals, and sharp images can be obtained.Moreover, since monochrome characters can be reproduced only in black, black characters can also be reproduced with high quality. Can be done.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing one embodiment of the present invention, FIG. 2 is a block diagram showing another embodiment of the present invention, and FIG. 3 is a block diagram showing an image signal from a monochrome original, which is the premise of the present invention. A system block diagram showing an example of an image processing device, FIG. 4 is a block diagram showing an example of the configuration of a dither processing circuit, and FIG. 5 is a block diagram for an area determination algorithm based on colored pixel density.
FIG. 6 is an explanatory diagram of an example of reference pixel arrangement, FIG. 7 is an explanatory diagram of a two-dimensional high-frequency emphasis filter used for MTF correction, and FIG.
Figure 9 is a circuit block diagram showing an example of a x3 filter, Figure 9 is a circuit block diagram showing an example of a colored pixel density calculation circuit, Figure 10 is a block diagram for an area determination algorithm based on edge pixel density. An explanatory diagram of an example of a differential filter, Fig. 12 is an explanatory diagram for finding the maximum value of gradients in eight directions, Fig. 13 is a block diagram for an area determination algorithm using halftone detection, and Fig. 14 is a pattern matching diagram. 15 is an explanatory diagram of a judgment correction process, FIG. 16 is a block diagram showing an example of a halftone dot detection circuit, FIG. 17 is a block diagram showing an example of an area determination circuit, FIG. 18 is a configuration diagram showing an example of a judgment correction circuit, and FIG.
The figure is an explanatory diagram of an example of a dark value matrix pattern for dither processing. t------One-person system 1.2-・--・Photograph processing section, 3.--11---Dot/text processing section, 4-.
---Area determination unit, 5--selector, 6--------
-Output system, 1-1-...-Color correction circuit, 1-3-...-
Halftone area determination circuit, 1-4.--Character area determination circuit,
1-5...-Color correction/UCR circuit, 1-6...
-.- Gradation processing circuit for halftone dots, 1-7...- Gradation processing circuit for characters, 1-8- Gradation processing circuit for knee photographs, 1-9.
1-10.1-13・--Selector, l-'11・-
・−・Output control unit, 1-12・−・・−m points・Character gradation processing circuit. 31st! ! (a) (C) Figure 4, Figure 5, Figure 6 (σ) (b)
(c)(d)
(e)
(f) Section 71i (a), (b) (c)
(d) (e) Scion Op ㅅ兔19 97 people Matrix (C Matrix Matrix (b) Matri 7-mata Ntrik people

Claims (1)

[Claims] In an image processing apparatus having a reading device that separates and reads a multicolor original into a plurality of colors, and an output device capable of outputting three or more values for each of the plurality of colors, for each image of each color. At least an image area discriminating means for discriminating into a density tone image area and an area tone image area, at least two types of dither processing means, and an optimal dithering process for the corresponding area according to the discrimination result of the image area discriminating means. selection means for selecting from the dither processing means, and when the image area discriminating means determines that even one of the plurality of colors is an area gradation image area, all color components are area gradation image areas. An image processing apparatus characterized by comprising a determination correction means.