JPH01252066A - Halftone picture separation processor - Google Patents

Abstract

PURPOSE:To obtain a picture with excellent separation and gradation by deciding whether a picture is in a binary picture area or a contrast picture area at every of plural picture element blocks depending on the mutual relation between the density gradation of a picture and the distribution state of the picture element and applying respectively a simple binarizing processing and dither processing respectively. CONSTITUTION:A/D conversion 12, shading distortion correction 13, AGC 14 are applied to an analog multi-level signal read by a CCD 11. The signal is stored respectively in memories 17, 18 while being binarized by a dither processing circuit 15 and a simple binary processing circuit 16. The black/white variable number A in the main and subscanning direction in the block is measured by the circuit 24, the density gradient A in the block is measured by the circuit 21, compared with specified values (k), (m) by comparators 26, 23 and the selector 28 is switched by OR 27 of the compared outputs. In the case of only A<k and A<m, the block is decided in the contrast picture area and the information of the memory 17 is given from the selector 28. In the case of A>=k or A>=m, it is decided to be in the binary picture area and the information of the memory 18 is outputted. By such a constitution, the original in mixture of character and photograph for its resolution and gradation is reproduced in an excellent way.

Description

Detailed Description of the Invention (Objective of the Invention) (Industrial Application Field) The present invention automatically separates areas of a document containing two images and a light side image, and divides the area into each image area. The present invention relates to a halftone image separation processing device that performs suitable processing.

(Prior Art) Conventional image processing devices that read a document with an image sensor and record or display it using a quantized image signal, when the document contains text and photographs, distinguish each part and process the image. Processing is being done. This is binarization for each pixel! 2! With LL, only text images are reproduced well, and there is a limit to the reproducibility of shading parts of photographs, etc. However, when pseudo halftone processing such as dither processing is performed, shading images of photographs, etc. are reproduced well. On the other hand, this is because the reproducibility of line drawings such as characters deteriorates.

For example, a case will be described in which the image information shown in FIG. 5 is subjected to dither processing.

In FIG. 5, a small square indicated by a broken line represents one pixel. Further, each large square indicated by a solid line represents one block 1 to 8, and each block is composed of 16 pixels. Needless to say, each block size may be any size of NXM (N, M are integers).

Further, the numbers shown for each pixel indicate the humidity level read from the original, with '0' indicating white, '15' indicating black, and levels in between indicating intermediate gray. In addition, the fifth
Although the density levels in the diagram are expressed as integers for convenience, they do not necessarily have to be integers.

To dither such image information, compare each block 1.2.3, 4.5, 6.7.8 shown in FIG. 5 with the dither matrix 9 shown in FIG. If the threshold value of is larger than the density level within the block, the pixel is determined to be white, and in the opposite case, the pixel is determined to be black. For example, when data 1.2, 3, 4.5, 6, and 7.8 shown in FIG. 5 are dithered, the processing results are as shown in FIG. 7.

When dithering is performed in this way, blocks 1.2.
It can be seen that the reproducibility of line drawing portions such as characters 5.6 is not good.

For this reason, there is a method disclosed in Japanese Patent Laid-Open No. 58-3374 as a method for distinguishing between character image areas and grayscale image areas. In this method, the document surface (screen) is divided into blocks each consisting of a plurality of pixels, and within each block, the difference in density level is determined between the pixel with the maximum density level and the pixel with the minimum density level.

If this difference is larger than a predetermined value, it is determined that it is a binary image O area or a character image part, and in other cases, it is determined that it is a light image area or a photographic image part.

For example, when the image information shown in FIG. 5 is processed by this method, the result will be as follows.

Now, assuming that the maximum value of the density level in the block is Pnax and the minimum value is Pnin, when Pnax - Piin≧m, binary image area (1) p
When nax −pnin <m, light image area (2) When m=8, blocks 1 to 8 are judged. FIG. 8 is a diagram in which blocks determined to be binary image areas are indicated by diagonal lines. That is, blocks 1 and 5 are blocks in the binary image area.

In this case, all blocks are divided into 2
The straight line in left block 1.5 appears clearly compared to the case of FIG. 7, which is converted into values. The two thin lines shown in block 2.6 on the - side are cut off, causing deterioration in image quality. Regarding this block 2.6, the amplitude density level of black and white is becoming smaller due to the MTF (modulation, transfer, function) characteristics of the reading device, so the density difference becomes smaller and the block is incorrectly determined to be a gray area. It is.

Therefore, if the m value in Equation (1) and Equation (2) is set to m = 6, the judgment regarding this block 2.6 will be in the binary image Q area, and the 1lIIl line will be reproduced well, but on the other hand, Block 4.7.8, which is a gray area, is also a binary image fIi
This will cause problems.

The diagram shown in FIG. 9 shows the output density level when scanning an image containing a mixture of line drawings and photographs.

A case will be explained using FIG. 9 in which a binary image area with a reduced MTF is erroneously determined to be a grayscale image. In the figure, B to B represent the pixel blocks, T1 to T5 represent the difference between the maximum and minimum pixel density within the block, and P represents the binary image fs!
This is a threshold value for determining whether it is a region or a grayscale image region.

Now, in FIG. 9, the density difference T2 of block B2 is the threshold value P
Since it is larger than , it is also clear from the graph that it is a binary image area. However, when looking at block B3, the graph shows that 2
Although it is considered to be a value image word range, since it is a fine pattern, the MTF decreases, and as a result, the m difference T3 becomes smaller than the threshold P, and it is erroneously determined as a light area. Therefore, if the value of the threshold P is set to be P smaller than the density difference ■3, block B3
is determined to be a binary image head area, but if this is done, in block B5, ■5 becomes larger than Pa even though it is a gray area, and it is incorrectly determined to be a binary image area. There was a problem with it going away. Therefore, since the value of the threshold P was compromised at an appropriate point between the two, erroneous judgments occurred in areas of shading images with relatively large changes in density, such as fine characters and line drawings, causing image quality deterioration.

In addition, this problem applies to all methods of determining and separating regions from density differences.
In the above device, the area is determined based on the density difference between pixels within a multi-pixel block, so when small characters or line drawings that reduce MTF appear, the density difference within the pixel block becomes smaller and is incorrectly determined as a dark and light area. Furthermore, the grayscale image area with relatively large density changes is a binary image fg! There was a problem in that it was incorrectly determined to be a distribution area. Therefore, in the present invention, this problem in the method of determining the image area by checking the density difference or density change is removed, and even if the MTF is reduced and the density difference or density change is small, such as fine characters or line drawings, binary images can be obtained. It is an object of the present invention to provide a halftone image processing device that correctly determines the area as a region, does not reduce the resolution of fine characters and line drawings, and performs good gradation reproduction even in the grayscale CLL area.

[Structure of the invention]

The present invention provides dividing means for dividing an input image screen into pixel blocks each consisting of a plurality of pixels, and a method in which the density gradient within the pixel block is larger than a predetermined value. Or, if the number of black and white changes in the main scanning direction and sub-scanning direction within the pixel block is larger than a predetermined value, the pixel block is determined to be a binary image area, and in other cases, the pixel block is determined to be a binary image area. It is characterized by comprising a determining means for determining the pixel block as a grayscale image area, and an image processing means for performing binary image processing and deep/dark image processing on both input items according to the determination result by the determining means. .

(Operation) In the present invention, in order to solve the problem that image block regions cannot be discriminated well based only on pixel density differences within blocks, other characteristics are also used to perform good image region discrimination.

The binarized image shown in FIG. 10 is an image obtained by binarizing the image shown in FIG.
It is used to explain the distribution state of valued pixels.

In FIG. 10, blocks 1, 2, and 5.6 are binary image areas, and blocks 3, 4, and 7.6 are gray image areas,
There are characteristics in the distribution of pixels that are binarized using constant values.

That is, when focusing on the number of changes in white pixels and black pixels within one block, there is a tendency for the number of changes in white and black to increase in a fine pattern in which the MTF decreases in a binary image creation area. On the other hand, in a gray-scale image area, since the density gradually changes, there is a tendency for one block to have one or no boundary line between white and black (the entire area is white or black).

Therefore, if we count the number of black and white changes M and N in the main scanning direction and the sub-scanning direction for each block and add them up for each block, we get the following.

Blocked... 8 times, Block 2... 12 times Block 3... 0 times, Block 4... 0 times Block 5...
- 8 times, block 6... 12 times block 7... 2 times, block 8... 5 times Next, divide this number into two groups using a certain number k as a threshold. Here, when =8 and M+N≧k, binary image area...(3) M+N<
When k, grayscale image area...(4), blocks 1.2, 5.6 are binary image areas, blocks 3, 4,
7.8 is determined as a grayscale image area.

Next, first calculate 2 from the density difference (density gradient) of pixels within the block.
Based on the results of grouping into value image areas and grayscale image areas, those that are grouped based on the distribution state of binarized pixels, and those that are determined to be binary image areas on both sides or on one side, as shown in Figure 4. is finally made into a binary image area, and the rest is made into a grayscale image area.

This means that even in a binary image area, for thick black lines or blocks at the boundaries of black areas with a large area, the face tree distribution state when binarized with a certain amount of value is different from the gray image area. This is because the concentration difference (density gradient) within a pixel block is given priority because it is indistinguishable from that of the pixel block.

In the present invention, it is determined whether each multiple pixel block is a binary image area or a grayscale image area based on the correlation between the image density gradient and the pixel distribution state, and each pixel block is converted into a binary image area by single@binarization. ! 2! Since dither processing is applied to the grayscale image area, it is possible to achieve binary images of mixed text and photos with good resolution and gradation, including fine text and fine line drawings where the MTF decreases. It is possible to obtain a processed output image.

(Example) An example of the present invention will be described in detail below based on the drawings. FIG. 1 is a block diagram showing the configuration of a halftone image separation processing VC device according to an embodiment of the present invention. As shown in the figure, this device is a charge-coupled device (CO
D) I L analog digital converter (A/D converter) 12, shading correction circuit (SHD correction circuit) 13, automatic gain control circuit (AGC) 14, dither processing circuit 15, simple binary processing circuit 16, memory 17.18 ,
Image signal data bus 19, intra-block concentration gradient measurement circuit 2
1. Threshold parameter holding circuit 22, comparator 23,
It consists of an intra-block pixel change measurement circuit 24, a threshold parameter holding circuit 25, a comparator 26, an OR gate 27, and a selector 28.

The CG[) 11 reads the original, and the read multilevel image signal is sent to the A/D converter 12. The A/D converter 12 converts analog Shintan into digital Shintan. S.H.
The D correction circuit 13 corrects shading distortion caused by variations in the amount of light from a fluorescent lamp as a light source. The AGC 14 automatically adjusts the gain of the image signal. The image signal whose gain has been adjusted is sent to the dither processing circuit 15 via the image signal data bus 19.
Simple binary processing circuit 16, intra-block humidity gradient measurement circuit 2
Sent to 1. The dither processing circuit 15 performs dither processing on the image signal. The simple binary processing circuit 16 performs simple binary processing of the image signal. Memories 17, 18 store both signals during block area determination.

The concentration gradient measurement circuit 21 within the block measures the concentration gradient within the block. The threshold parameter holding circuit 22 is based on the formula (1
), (2) threshold m is stored. Comparator 2
3 compares the output value of the concentration gradient measurement circuit 21 in the block and the output threshold value m of the threshold parameter holding circuit 22, and if the concentration gradient in the block is large, outputs 1", and if it is small, outputs "0". Intra-block pixel change measurement circuit 2
4 is from white pixel to black pixel in the main scanning direction, JJ
The total number of changes in both scanning directions is measured. The threshold parameter holding circuit 25 stores the threshold K of equations (3) and (4). The comparator 26 compares the intra-block pixel change measuring circuit 24 and the threshold output from the threshold parameter holding circuit 25, and outputs "JlPl" if the change in black and white within the block is greater than the threshold;
O'' is output.The OR gate 27 outputs the comparator 23,
26 output values are logically summed. The selector 28 outputs the value stored in the memory 18 if the output value of the OR gate 27 is 1'', and outputs the value stored in the memory 17 if the output value of the OR gate 27 is O''.

Next, the operation of this embodiment will be explained. The analog multivalued image signal read by the CCD 11 is converted into a digital image signal by the A/D converter 12. and,
A 5t-10 correction circuit 13 corrects shading distortion due to variations in light i of a fluorescent lamp as a light source, and an AGC circuit 14 adjusts the gain of the image signal.

The image signal read by Ccoii in this way is corrected in - different ways. The image signals are each binarized by the dither processing circuit 15 and the simple binary processing circuit 16, and are temporarily stored in the memories 17 and 18 while the block area is being determined, and when the block area is determined, According to the result, it is selected by the selector 28 and output as a binary image signal. On the other hand, for the simply binarized aunt, the intra-block pixel change measuring circuit 24 measures the total number of times the pixel changes to white within the block in both the main scanning direction and the sub-scanning direction, and the value is stored in the threshold parameter holding circuit. The comparator 26 compares the value of the threshold value outputted by the block 25 with the value of the threshold value outputted by the comparator 25.
On the other hand, the intra-block concentration gradient measurement circuit 21 measures the concentration gradient within the block, and the comparator 23 compares the value with the value of the threshold m output from the threshold parameter holding circuit 22. , if the concentration gradient within the block is large, "1" is output, and if it is small, "0" is output. Next, the outputs from these comparators 23 and 26 are logically summed by an OR gate 27.

If the output value of OR gate 27 is ``1'', selector 2
8 outputs the value stored in the memory 18, and if the output value of the OR gate 27 is 0'', the selector 28 outputs the value stored in the memory 17.
Outputs the value stored in .

Therefore, in this embodiment, the comparator 23 compares the gradient e in the block with a predetermined threshold value m, and the comparator 26 compares the number of pixel changes in the block with a predetermined threshold value m. , 26, the selector 28 is switched. Therefore, only when the density gradient shown in FIG. 4 is smaller than a predetermined threshold value and the number of changes between black and white is smaller than a predetermined threshold value m, the block is determined to be a gray area and is dithered by the dither processing circuit 15. Both signals are output from selector 28. On the other hand, if the density gradient is larger than the predetermined threshold IaK or the number of black and white changes is larger than the predetermined threshold m, the block is determined to be a binary image area, and the simple binary processing circuit 16 determines that the block is a binary image area. simple 2
The value-processed image signal is output from the selector 28.

FIG. 2 is a diagram showing processing results according to this embodiment.

As shown in Figure 2, blocks 1, 2, and 5゜6 1111M on the left half are clearly reproduced, and blocks 3゜4 on the left half are clearly reproduced.
．． 7.8 density image dither processing is performed, and good discrimination reproduction can be performed even for thin line portions where the MTF decreases, which could not be reproduced conventionally.

In this way, according to this embodiment, it is possible to simultaneously perform effective discrimination on originals containing binary images that require high resolution, such as character images, and gray and white images that require gradation, such as photographs. Binarization processing suitable for each image can be performed, and the resolution and gradation of originals containing text and photographs can be reproduced well.

In addition, parts of fine characters and thin line patterns that were previously erroneously determined to be grayscale image areas are now accurately determined to be binary image areas. Since the binary image @ area and the grayscale image area are discriminated at the same time, it is possible to switch to a redundancy suppression encoding method suitable for the characteristics of each image.

Therefore, if this device is used as a document reading system, even documents containing text and photographs can be automatically scanned for two images that are suitable for the nature of each image.
Since digitization processing is performed, the image quality of digital copies can be improved. Furthermore, if used in a facsimile reading system, it eliminates the need to switch between text and photo modes by pressing a button, and also has the advantage of being able to use mixed originals, making it effective for transmitting all types of image information.

Note that the present invention can be modified in various ways within the scope of its technical idea.

That is, in this embodiment, the density gradient in a pixel block was investigated using the density difference disclosed in Japanese Patent Laid-Open No. 58-3374, but since it is sufficient to know the tendency of the density gradient, other methods may be used. The concentration gradient may be binarized. For example, the density gradient may be determined by examining the distribution of density levels within a pixel block and then binarized.

In addition, in the density level binarization method, the binarization process for both CC and C image areas uses Bayer's threshold matrix,
Furthermore, although we have explained the binarization process using a single threshold value for the character image e area, it goes without saying that the binarization process after image area determination can also be performed using other methods. do not have. For example, Baver's threshold matrix is used in processing the grayscale image region, but various threshold arrays such as those shown in FIGS. 3(a), (b), and (c) may also be used. Furthermore, in the character image area, although simple binarization using a single threshold value is performed in the present invention, binarization may be performed using the Nwa rib slice method, contrast enhancement, or the like.

Also, even in grayscale images (photographs), blocks that have many densities that are biased toward white or black may be determined to be binary image areas, but since the blocks originally emphasize contrast, they are determined to be binary image areas. Even if you process the image, it will have almost no effect on the image quality.

Also, in the above explanation, the block size is composed of 16 pixels of 4 x 4 and has 16 gradations, but if this block size is configured as 8 x 8 or 8 x 4, it is possible to create 64 gradations or 32 gradations. Gradation reproduction may also be performed.

114 when the judgment block size is set to 8x8 or 8x4
The ili dither matrix corresponds to 8×
Although it is preferable to use 8 or 8×4, it is not necessary.

(Effects of the Invention) As described in detail above, according to the present invention, fine characters with reduced MTF and small density differences and density changes can be correctly determined as binary image areas even in line drawings. , it is possible to perform good gradation reproduction even in a dark and dark image area without reducing the resolution of line drawings.

[Brief explanation of the drawing]

FIG. 1 shows a medium 12XIWAiIi according to an embodiment of the present invention.
A block diagram showing the configuration of the i-image separation processing device, Figure fJ52 is a diagram showing the processing results of this embodiment, Figure 3 is a diagram showing another dither matrix, Figure 4 is a diagram showing area determination based on density gradient and pixel change. FIG. 5 is a diagram showing image information to be processed,
Fig. 6 is a diagram showing a dither matrix, Fig. 7 is a diagram showing image information processed by the dither method, and Fig. 8 is a diagram showing the image information processed by the dither method.
8-3374, a diagram showing the results processed by the method disclosed in No. 8-3374, No. 9 is a diagram showing the diagram output density level,
FIG. 10 is a diagram showing processing results when binarization processing is performed using uniform threshold values. 15...Dither processing circuit, 16...Simple binary processing memory, 17.18...Memory, 21...Ω in block
degree gradient measuring circuit, 22.25...threshold parameter holding circuit, 23.26...comparator, 24...internal pixel change measuring circuit, 27...OR gate, 28
···selector. (CI) (b)
(C) Figure 3 Figure 4 order m-page procedure manual (method) Mr. Commissioner of the Patent Office 1986
July 13, 2015 2, Name of the invention Halftone image separation processing device 3, Relationship with the person making the amendment Patent applicant (307) Toshiba Corporation 4, Agent (104) 2-chome, Ginza, Chuo-ku, Tokyo 11-2 Ginza Daisaku Building 6th floor Telephone 03-545-3508 <Representative)
Starting point June 28, 1988) 7. Contents of amendment (1) "No. 9 is a figure" on page 20, line 7 of the specification of the present application
is corrected to [Figure 9].

Claims (1)

[Scope of Claims] Dividing means for dividing the screen of an input image into pixel blocks each consisting of a plurality of pixels; If the number of black and white changes in the direction and sub-scanning direction is greater than a predetermined value, the pixel block is determined to be a binary image area, and in other cases, the pixel block is determined to be a grayscale image area. 1. A halftone image separation processing device, comprising: a determination unit for performing binary image processing and a grayscale image processing on an input image according to a determination result by the determination unit.