JPH0446500B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an image identification method for classifying and identifying an image into each area of a line drawing, a halftone image, and a continuous tone image.

(Prior Art) As a conventional image identification method, for example, the method shown in the following document is known.

Mitsuhiro Murao, Toshifumi Sakai “Extraction of structural information from document images” 1980, Information Processing Society of Japan 21st National Conference 7H-1 Morizumi Kurose, Koichi Ejiri, Akira Sakurai “Real time of halftone images in document images Identification “1980, Information Processing Society of Japan 23rd National Conference 6C-6 Hiroshi Makino, Yuku Akada “Binary Reproduction of Document Images Containing Dark and Light Regions” March 1980, Theory of IEICE (D) J65-D ,
3, pp.307-314 Kenichi Takahashi, Masamitsu Ota “Computer-based Image Classification for Compression Coding” August 1980, IEICE Theory (D)
J65-D, 8, pp.1018-1025 JP-A-56-132061 “Halt image signal detection method” Hiroshi Ueno, Takayoshi Semasa “Tone reproduction in newspaper grid” Print magazine, 1982 (Vol.66) )1, pp.15
~23 JP-A-56-149674 “Method for identifying image characteristics” (Problems to be solved by the invention) These conventional methods are used to identify whether a line drawing or a continuous tone image, or whether a line drawing or a halftone image The identification was the main one. By the way, in reality, most images are a mixture of line drawings, halftone gradation drawings, and continuous gradation drawings, and the above three
A way to identify the species is also needed. From this point of view, the present inventor conducted a study on image identification methods in the Showa era.
3 above at the 1167 National Conference of the Institute of Electronics and Communication Engineers in 1958.
We proposed a method for identifying types. However, this method has the problem that it is easy to misidentify continuous tone images with rapid density changes.

The present invention has been made in view of this problem, and its object is to provide an image identification method that can accurately discriminate between halftone gradation images, line drawings, and continuous tone images.

(Means for Solving the Problems) The present invention to solve the above problems binarizes an image, divides it into a plurality of identification unit areas, and calculates the number of black pixels constituting a black pixel area within the identification unit area. If the variation is small, it is determined that the discrimination unit area is a halftone gradation image, and the proportion of the background within the discrimination unit area is determined. If the proportion is large, the discrimination unit area is determined to be a halftone image. The identification unit area is determined to be a line drawing, and if the discrimination unit area is determined to be neither a halftone gradation drawing nor a line drawing, the identification unit area is determined to be a continuous gradation drawing.

(Example) Hereinafter, an example of the method of the present invention will be specifically explained using the drawings.

In the method described here, an image is first divided into a plurality of identification unit areas (minimum units for determining whether it is a line drawing, a halftone gradation image, or a continuous gradation image). For example, when the image is N×M=2000×2500 pixels (10 dots/mm) as shown in FIG. 1, the size of the identification unit area is set to k×k=32×32 pixels. Next, the image is binarized, and a black pixel area is determined pixel by pixel for each identification unit area. Here, one black pixel area is
For example, at least one of four directions: up, down, left, and right.
It is a set of black pixels that are adjacent to each other in the direction, and if there is no black pixel adjacent to one black pixel, a single black pixel forms one black pixel area. Note that the black pixel referred to here refers to a white pixel in a black-and-white inverted image. FIGS. 2A and 2B show an example of binarizing a halftone gradation image and a continuous gradation image, respectively (parallel vertical lines (pair) in the figures indicate black pixels). In the case of halftone gradation painting,
It can be seen that although the variation in the number of black pixels constituting each black pixel area is small, the variation is large in the case of continuous tone images. Although not shown, there are large variations in line drawings as well. Focusing on this point, the method of the present invention determines the variation (standard deviation, etc.) in the number of black pixels constituting each black pixel area for each identification unit area, and determines whether it is a halftone gradation image or not. There is. Although not as great as the variation in the number of black pixels, the variation in the distance between black pixel areas (standard deviation, etc.) also differs between halftone gradation images and continuous gradation images. Figure 3 shows the black pixel area A ₁ ,
This shows an example of distance measurement positions when calculating the distance between A ₂ and indicates that the distance between the centers of gravity of each black pixel area A ₁ and A ₂ is measured. FIG. 4 shows the standard deviation of the number of black pixels within the black pixel area and the standard deviation of the distance between the black pixel areas obtained in this manner, with the former taken as the vertical axis and the latter as the horizontal axis. In the figure, the black circles are the results obtained for some of the discrimination unit areas of the halftone gradation image, and the black squares are the results obtained for some of the other discrimination unit areas. In the case of halftone gradation images, it can be seen that the standard deviation of the number of black pixels is smaller than in other cases, and this tendency remains the same even when many samples are used. Therefore, in the method of the present invention,
Whether or not the corresponding identification unit area is a halftone gradation image is determined based on whether or not the standard deviation of the number of black pixels, in particular, is smaller than a predetermined value. For example, if the standard deviation of the number of black pixels is within 1.0 pixels, it is determined that the image is a halftone gradation image. Note that this determination can also be made using the standard deviation of the distance between black pixel areas, but as is clear from Figure 4, the standard deviation of the distance between black pixel areas also Since there are cases where the distance between black pixel areas is small, accurate determination cannot be made only by the standard deviation of the distance between black pixel areas.

Next, in the method of the present invention, the proportion of the background is examined. This ratio is proportional to the number of white pixels in the discrimination unit area because the size of the discrimination unit area is fixed, so it can be easily calculated by finding the total number of white pixels or the total number of black pixels in the discrimination unit area. . but,
The total number of white pixels or the total number of black pixels may be used as data indicating the ratio of background pixels without actually calculating it. FIG. 5 shows the proportion of this background, and the horizontal axis shows each sample in the identification unit area. In the figure, the black circles are the identification unit areas of continuous tone drawings, and the black squares are the case of line drawings. In the case of line drawings, since the proportion of the background is large, the difference is large compared to continuous tone drawings. Therefore, it is easy to determine whether an image other than a halftone gradation image is a line drawing or not.

By the above method, line drawings, halftone gradation drawings, and continuous gradation drawings can be identified.

The device that performs this identification includes, for example, a memory for storing various data such as image data, a means for calculating the black pixel area in each identification unit area and the number of black pixels in the black pixel area, and means for determining the variation in the number of black pixels of the area; first discriminating means for determining that the discrimination unit area is a halftone image when the variation is small; and means for determining the proportion of the background within the discrimination unit area. , if the ratio is large, a second discrimination means discriminates that the discrimination unit area is a line drawing, and the discrimination unit area that is not discriminated as either a halftone gradation image or a line drawing by the first and second discrimination means is continuous. and a third discriminating means for discriminating that the image is a gradation image. Among the above-mentioned means, the means other than the memory are usually constituted by a computer in terms of hardware.

By the way, the above explanation is for the case where the discrimination is performed over the entire area of the image, but if the image area is extracted, the above method may be applied to each image area. In this way, the area range to be identified becomes narrower, so that efficient identification can be performed.

Therefore, an example of a method for extracting an image area will be described next. FIG. 6 is a flowchart showing the steps in a specific example of this extraction method; in this example,
First, the background level (density level) is determined, then the image is divided into blocks, the pixels in each block are binarized using a threshold value determined based on the background level, and the texture analysis of each block is performed from the binarized image data. Each block is classified into a predetermined pattern, and the connections between the blocks are examined using the pattern to extract similar image areas. Note that the order of background level determination and image blocking may be reversed.

Next, the above procedure will be explained in detail step by step.

() Background level determination In order to determine the background level in a short time, m x m (for example, 32
×32) Create a pixel window,
A density histogram as shown in FIG. 8 is created based on the image data in the window, and the background level is determined. FIG. 9 is a flowchart showing the procedure for determining the background level. First, a window is determined (position determined), a density histogram is created based on the window, and the average value M and standard deviation S are calculated. Next, the standard deviation S is set to a predetermined threshold T
If it is smaller, the background level B is selected as a value within the range of M-3S≦B≦M+3S. Usually, B=M+
3S. On the other hand, if the standard deviation S is equal to or greater than the threshold value T, the window position is moved in the horizontal or vertical direction in FIG. 7 to determine a new window, and the density histogram is created again. This is because when S≧T, the window does not exist at a position where the background can be captured well.
Note that the size of the threshold T may be any value as long as it is within a range that can capture the background, but it is preferably a size that allows the window to be determined a minimum number of times.

() Blocking an image Divide an image into multiple blocks. For example, when the image is N×M=2000×2500 pixels (10 dots/mm) as shown in FIG. 10, the size of each block is set to n×n=32×32. Of course, 16×16 or 64×
You can also choose 64 etc.

() Texture analysis of each block Next, the image data is binarized using the background level B as a threshold, and each block is classified into a predetermined pattern using the binarized image data obtained thereby. As the patterns prepared in advance, for example, 16 patterns 0 to G as shown in FIG. 11 are used. This pattern changes as it changes from 0 to G, i.e. (0,1) → (2~5) → (6~
The pattern has many black and white changes each time it changes from 7) to (8 to B) to (C to F) to G. To classify each block into the above patterns, for example, as shown in FIG.
Pattern classification can be easily performed using a computer or the like by using a method of determining the cumulative number of , and determining the corresponding pattern from the distribution of these two cumulative numbers in the X and Y directions. In the example in Figure 12, black pixels are concentrated at the bottom right of the block, so
In the direction, the cumulative number on the right side is large, and in the Y direction, the cumulative number on the bottom side is large, and
Since pattern 8 also shows the same tendency when projected in the two-dimensional direction, it is concluded that the block corresponds to pattern 8.

() Extraction of image region The connection between each block may be examined in units of blocks, or each block may be divided into two vertically and horizontally (four divisions in total) and examined.
Figure 13 is a simplified explanatory diagram to show the latter method.
Figure 13 (b) shows the binarized image superimposed on it (the shaded area is the black pixel area).
Figure 13 (c) with the pattern numbers in the figure shows how each block is divided into four as described above (for example, block α → small blocks ₁ , α ₂ , α ₃ , α ₄ ), and each of the obtained small blocks is divided into four. FIG. 12 is a diagram in which the pattern numbers of FIG. 11 are assigned. When each pattern in FIG. 11 is divided into four, it is possible to know in advance which patterns each pattern will be decomposed into, so it is extremely easy to obtain FIG. 13(b) to FIG. 13(c). still,
When each pattern in Figure 11 is divided into four, the first
Although each pattern in FIG. 1 may be selected, it is sufficient that the patterns can be approximately decomposed into the patterns in FIG. 11. Next, the connections between the small blocks are examined, and in order to determine that there is a connection relationship, for example, the criteria for determining whether there is a connection relationship is that there is at least a black part on the joining edge of patterns of adjacent small blocks. Alternatively, the presence or absence of a connection relationship may be determined by taking into consideration the overall shape of the pattern. In any case, since the combinations of pattern connections are limited, it is possible to determine whether or not small blocks are connected by determining in what positional relationship each pattern should be adjacent to each other to determine if small blocks are connected. The decision is quite simple. Figure 13 D is a diagram showing an image area obtained by assuming that there is no connection relationship between small blocks when there is only a white part on the joint side of the pattern. Areas with the same number in the figure are image areas of the same type. be. Of course, the connections between blocks may be checked using the above method without performing the four-division process. However, in this case, the boundaries of the regions become rough. In order to solve this problem of roughness, it is conceivable to reduce the size of the block from the beginning, but this would require a considerable amount of time for the texture analysis.

As a result of the above, the image area has been extracted. Figures 14 to 16 are more specific explanatory diagrams of area extraction, with Figure 14 showing the original image, Figure 15 showing the pattern classification image, and Figure 1
FIG. 6 is a diagram showing a region extracted image. Furthermore, the 14th
The areas indicated by 1 and 2 in the figure are filled with characters, and the areas indicated by 3 and 4 are continuous tone images. As shown in FIG. 15, the image extraction is successful, but in FIG. 15, there is a raised area (block) due to dirt etc. in the original image, and this creates a unique image area in FIG. 16. ing.
In order to prevent this noise, processing may be performed to treat the raised block as being in the same image area as the surrounding blocks.

After extracting image regions as described above, by applying the above-described identification method to each extracted image region, efficient image identification can be performed.

(Effects of the Invention) As explained above, in the method of the present invention, halftone images are determined based on the variation in the number of black pixels, line drawings are determined based on the proportion of the background, and the rest are determined as continuous tone images. , halftone gradation drawings, line drawings, and continuous gradation drawings can be accurately distinguished.

[Brief explanation of the drawing]

Figure 1 is an explanatory diagram showing an example of the identification unit area, Figure 2
The figure is an explanatory diagram of a binarized image, Figure 3 is an explanatory diagram of distance measurement between black pixel areas, and Figure 4 is a standard deviation of the distance between black pixel areas and the standard of the number of black pixels constituting the black pixel area. Figure 5 is a diagram showing the relationship with deviation, Figure 5 is a diagram showing the ratio of background, Figure 6 is a flowchart showing the procedure in a specific example of the area extraction method, Figure 7 is an explanatory diagram of the window, and Figure 8 is Explanatory diagram of density histogram,
Fig. 9 is a flowchart showing the background level determination procedure, Fig. 10 is an explanatory diagram of image blocking, Fig. 11 is an explanatory diagram of block patterns, Fig. 12 is an explanatory diagram of pattern classification, and Fig. 13 is an illustration of the pattern classification. is an explanatory diagram of image region extraction, FIG. 14 is a diagram showing the original image, FIG. 15 is a diagram showing a pattern classification image,
FIG. 6 is a diagram showing a region extracted image. α...Block, _α1 , _α2 , _α3 , _α4 ...Small block,
1, 2...Character image area, 3, 4...Continuous tone image area.

Claims (1)

[Claims] 1 While binarizing the image, divide it into multiple discrimination unit areas, find the variation in the number of black pixels that constitute the black pixel area within the discrimination unit area, and if the variation is small, the discrimination unit area is In addition to determining that the identification unit area is a toned image, the ratio of the background within the identification unit area is determined, and if the ratio is large, the identification unit area is determined to be a line drawing, and based on the above determination, it is determined whether it is a halftone gradation image or a line drawing. If not, an image identification method that determines that the identification unit area is a continuous tone image.