JPS6358427B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to an image processing method, and particularly to an image processing method for distinguishing between a black and white binary image such as a character image and an image having photographic halftones. Generally, in digital image processing, it is not desirable to use one image processing method for all types of images, such as character images, line images, photographic images, and the like. Even if the results are good for some types of images,
This is because other types of images often have poor results. For example, a dither method using a so-called dither matrix for converting a multivalued image into a binary image is suitable for processing photographic images, but is inappropriate for document images. Therefore, if the entire image is of one type, it is possible to select from the beginning whether or not to use processing such as the dither method. However, when there are multiple types of images in one image, it is necessary to automatically sort out the types and decide what processing to perform. The present invention has been made based on the above circumstances, and an object of the present invention is to provide an image processing method that can identify images having photographic halftones in real time. To achieve this objective, the present invention reads an image pixel by pixel, assigns a first numerical value to pixels at halftone density level, and assigns a second numerical value to pixels at other density levels, and converts the image into a binarized image. , the calculation of the run length of the pixel having the first numerical value is performed using the first numerical value.
Regarding the main scanning direction (X-axis direction) of the pixel of interest having the numerical value of , for a total of 9 consecutive pixels, 4 on each side of the pixel of interest, and the sub-scanning direction of the pixel of interest having the first numerical value ( (Y-axis direction) for a total of 9 consecutive pixels, 4 on each side of the pixel of interest, and when the calculated value is 4 or more in both the main scanning direction and the sub-scanning direction, the above The present invention is characterized in that the pixel of interest is determined to be a photographic halftone. As a result, when the pixel of interest is a halftone pixel, the present invention can
The run length of a pixel with halftone density in both directions of the axis is calculated, and when the run length exceeds a predetermined threshold value in both directions, the pixel of interest is determined to be part of a photographic image. Therefore, according to the present invention, images with photographic halftones and images without photographic halftones can be distinguished in real time, and pixels with halftone densities in both the X-axis and Y-axis directions centered on the pixel of interest can be identified. By determining the run length of the image, it is possible to reliably distinguish between halftone images that are created when printed characters are crushed or smeared, and halftone images in the original sense such as photographs. Furthermore, since the halftone discrimination is performed for each pixel, precise image reproduction is possible. Embodiments of the present invention will be described below with reference to the accompanying drawings. First stage According to this invention, as a conversion function F(x) where x is the concentration level, when α<x<β, F(x)=1, when x>α or x<β, F(x)= Binary quantize the input image using a function such as 0.
That is, the intermediate density level is output as "1" and the other density levels are output as "0". In this case, if the input image is expressed with 8 bits = 256 levels, the input maximum density level is
255, it is desirable to set α=250 and β=about 80 to 100. This process is based on the following facts. In other words, the density frequency distribution in character images and line images is as shown in Figure 2 a, and in photographic images, it is as shown in Figure 2 b.
It looks like this. As can be seen, in character images and line images, there are few pixels at intermediate density levels, and they are concentrated only at the edge portions of the image. In contrast, in photographic images, there is no extreme variation in each density level. However, according to this step, only the edge portions of character images and line images are extracted. Second Step Calculate the run length from the surrounding pixels of the pixel of interest in the image binarized in the first step. and,
If this run length is greater than or equal to a certain threshold, it is determined that the pixel of interest is a halftone in a photographic image with many intermediate density levels (that is, a photographic halftone).
This is based on the fact that document images and the like have extremely short run lengths, but photographic images and the like generally have long run lengths, as explained with reference to FIG. The run length is calculated as follows. (1) When the level of the pixel of interest X is X=0, this pixel of interest is not judged and processing moves on to the next pixel. (2) When the level of the pixel of interest X is X=1, a mask as shown in FIG. 3 is set in the main scanning direction (left and right) and the sub-scanning direction (up and down) of the pixel of interest X. That is, n pixels A ₁ to _{A o} are placed to the right of the pixel of interest X, and n pixels A _-1 to _{A -o} are placed to the left of the pixel of interest.
, n pixels B ₁ to _{B o} are taken in the upper direction, and n pixels B _-1 to _{B -o} are taken in the lower direction. In this case, if the run length threshold is, for example, 4, then n=4. Here, level “1” is continuous in the main scanning direction.
The number of is counted as the run length value. For example, if A _-o・…・A _-2・A _-1・X・A ₁・A ₂・…・B _o 1・…・0・1・1・0・1・…・0, then run The length value is 2. Similarly, the run lengths of pixels B _-o to _{B o} in the sub-scanning direction are calculated, and if both run lengths exceed the threshold, it is determined that the image has photographic halftones and outputs "1"; If only one of them exceeds the threshold, or if both of them exceed the threshold, a determination of "0" is output. Generally, a run length threshold of about 4 or 5 is sufficient. Next, a specific configuration for executing the above processing will be described with reference to FIG. 4. According to the figure, a binary converter 11, a demultiplexer 12,
Nine input buffer registers 13A to 13I, multiplexer 14, buffer register 15, reference pattern generator 16, comparator 17, register 1
8, and AND gate 19, 4 bits=
A case is shown in which 16 levels of input pixels are processed with a run length threshold of 4. The binary converter 11 is a ROM having conversion contents as shown in Table 1 below.

[Table] As can be seen from Table 1, in this example, α=13, β=
You can see that it is set to 5. The output data of this converter 11 is sequentially input to nine input buffer registers 13 (13A to 13I) via a demultiplexer 12. Each input buffer register 13 can store data for one scanning line. Also, set the run length threshold to 4
Then, n = 4 in Figure 3, and to calculate the run length of 9 pixels in the vertical and horizontal directions, 9 scanning lines are required.
This would require several registers. These buffer registers 13A to 13I are
The virtual address can be converted when processing of a certain scanning line is completed. That is,
If the pixel of interest is in the register 13E at a certain point, the address is set to 1 when processing the next scanning line.
3A → 13B, 13B → 13C..., 13G → 1
Convert as follows: 3H, 13H → 13I, 13I → 13A. Among the contents of such buffer registers 13A to 13E, the nine pixels necessary to calculate the run length (for example, in FIG. 3, A _-4 to X to
A ₄ ) are sequentially accumulated in the buffer register 15 via the multiplexer 14. This buffer register 15 is a register having at least nine serial-in and serial-out stages. On the other hand, the reference pattern generator 16 includes a ROM, and stores all patterns that may be formed by combinations of nine pixels.
For example (000010000), (000011000), (000011100)
..., (101011100). However, some of the stored codes of the reference pattern generator 16 always match the contents of the buffer register 15. Therefore, buffer register 1
A comparator 17 sequentially compares the contents of 5 with the contents of the reference pattern generator 16, and when they match, the comparator 17 selects a predetermined signal So. That is, according to this embodiment, since the run length value is 4, (000011110), (00011100)...
..., (111111111), the comparator 17 outputs a judgment "1" for a code with four or more consecutive 1's,
In other cases, the comparator 17 is configured to output a judgment of "0". Similarly, the run length value in the sub-scanning direction is calculated, and the same determination as the previous determination of the run length value in the main scanning direction is made. However, the output of the comparator 17 is sequentially input to a two-stage shift register 18. This shift register 18 is a two-stage serial-in/parallel-out register, and the contents of each stage are input to an AND gate 19. Therefore, when the output of the comparator 17 is continuous (1, 1), the AND gate 19 outputs So=1, and otherwise outputs So=0. In this way, when So=1, it is determined that the image has a prominent intermediate density level, and when So=0, it is determined that the image is not. As explained above, according to the present invention, when the pixel of interest is a halftone pixel, the run length of the pixel of halftone density in both the X-axis and Y-axis directions centered on the pixel of interest is calculated, and When a predetermined threshold value is exceeded, the pixel of interest is determined to be part of the photographic image, so images with photographic halftones can be identified in real time, and the pixel of interest can be identified as a part of the photographic image. Since the run length of pixels with halftone density in both the X-axis and Y-axis directions is determined, it is possible to distinguish between halftone images that are created when characters are crushed or blurred when printed, and the original meaning such as a photograph. The excellent effect of being able to reliably distinguish between the halftone image and the halftone image can be obtained.
Further, since the halftone identification is performed for each pixel, there is an advantage that accurate image reproduction is possible.

[Brief explanation of the drawing]

Fig. 1 is an explanatory diagram showing functions for performing binarization processing according to the present invention, Fig. 2 is a diagram showing density frequency distribution of various images, and Fig. 3 is an explanatory diagram showing a function for performing binarization processing according to the present invention. FIG. 4 is an explanatory diagram of an embodiment of the present invention. 11... Binary converter, 12... Demultiplexer, 13... Input buffer register, 14... Multiplexer, 15... Buffer register, 16
...Reference pattern generator, 17...Comparator, 18
...Shift register, 19...And gate.

Claims (1)

[Claims] 1 Read the image pixel by pixel, assign a first numerical value to the pixels at the halftone density level and a second numerical value to the pixels at other density levels, and binarize the image, and in the binarized image, The run length of a pixel having a numerical value of 1 is calculated for a total of 9 consecutive pixels, 4 on each side of the pixel of interest having the first numerical value, with respect to the main scanning direction of the pixel of interest having the first numerical value, and 4 on each side of the pixel of interest in the sub-scanning direction, with a total of 9 pixels.
The image is characterized in that the pixel of interest is determined to be a photographic halftone pixel when the calculated value is 4 or more in both the main scanning direction and the sub-scanning direction. Processing method.