JPS6135673A - Area extracting method - Google Patents

Abstract

PURPOSE:To extract the picture area of the same kind by dividing a picture into plural blocks, using binary-coded picture data, classifying each block into a predetermined pattern and checking the connection between the blocks. CONSTITUTION:A picture element window is provided on a picture area to decide a background level in a short time, a density histogram is generated based on the picture data in the said window to decide the background level. Then the picture is divided into plural blocks. Then the picture element in each block is binary-coded by a threshold value obtained based on the background level, texture analysis of each block is applied from the binary-coded picture data and each block is classified into a predetermined pattern. Then the picture region of the same kind is extracted by checking the connection between the blocks. Further, the order of the decision of background level and the blocking of the picture is replaced.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a region extraction method for extracting similar image regions from an image.

(Prior Art) As this type of region extraction method, the 12th Image Engineering Conference 7-2 "Recognition of Rectangular Fields and Handwritten Identification Numbers in Facsimile Documents" is known. In this method, the image is first divided into small areas, then 2m wide, a logic plate is used to combine the entire screen into multiple large rectangular areas, and then each large rectangular area is divided into window areas. Find the pixel occurrence probability and the local bias within the black pixel window. Furthermore, the area projection of the data onto the coordinate axes (X, Y) is determined as the marginal distribution.

Then, based on the probability of occurrence of black pixels and the local bias within the window, we looked at the attributes of ``photo J,''``diagram,'' and ``text,'' and examined the periodicity of the zero distribution. ” and performs region extraction.

(Problem that the invention purports to solve) By the way, this method first has a poor correspondence of λ = 1 in an arbitrary shape region in order to group it into a human figure region, and also for character units without periodicity. The problem is that region extraction is difficult. Therefore, a method that is adaptable to arbitrary shapes and arbitrary images has been desired.

The present invention has been made in view of the above problems, and an object of the present invention is to provide a region extraction method capable of extracting an image region of an arbitrary shape.

(Means for Solving the Problems) The present invention, which solves the above problems, divides an image into a plurality of blocks, classifies each block into a predetermined pattern using binarized image data, and This method is characterized by examining the connection between each block and extracting image regions of the same type.

(Example) Hereinafter, the method of the present invention will be specifically explained using the drawings. FIG. 1 is a diagram showing the steps in a specific example of the method of the present invention.
- In this example, the background level (mlt degree level) is first determined, then the image is divided into blocks, and the pixels in each block are divided by 2 using the PI value determined based on the background level.
Each block is digitized and texture analyzed from the binarized image data, each block is classified into a predetermined pattern, connections between each block are investigated using the pattern, and similar image areas are extracted. Note that the order of determining the background level and dividing the image into blocks may be reversed.

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

(I> Determination of background level In order to determine the background level in a short time, a window of (n+xm (for example, 32x32) pixels) as shown in Figure 2 is provided on the image area, and the image data within the window is Based on this, a 11111 LCE histogram as shown in Fig. 3 is created and the background level is determined. Fig. 4 shows the procedure for determining this background level.
First, a window is determined (position W is determined), an m-degree histogram is created based on the window, and the average value M and standard deviation S are calculated. Next, if the standard deviation S is smaller than a predetermined threshold T, the background level B is selected as a value within the range of M-38≦B≦M + 38. Usually, B = M + 38
shall be. On the other hand, if the 1M standard deviation S is 1ltlT or more, the window position is moved in the horizontal or vertical direction in FIG. 2, a new window is determined, 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. The size of 1lliIT may be any value as long as it can capture the background, but it is preferably such a size that the window can be determined a minimum number of times.

(II) Blocking the image Divide the image into a plurality of blocks. For example, if the image is NxM=2000x2500 pixels (10dots) as shown in Figure 5,
/ll1m), the size of each block is n x
Take n = 32x32.

Of course, you can also choose 16X16, 64X64, etc.

(I) 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 for the patterns prepared in advance, for example, 16 patterns (0
~G) is used. This pattern changes as it changes from O to G, i.e. (0,1) → (2~5) → (6,7) → (
The pattern has many black and white changes during the transition from 8 to B) → (C...F) → Q. For example, as shown in FIG. 7, the classification of each block into the above patterns is
Black pixels (
Pattern classification can be easily performed using a computer, etc. by calculating the cumulative number of "1°°) and finding the corresponding pattern from the distribution of these two cumulative numbers in the XYh direction.The example in Figure 7 In this case, since the black pixels are concentrated at the bottom right of the block, the cumulative number on the right side is large in the X direction, and the cumulative number on the bottom side is large in the Y direction, and pattern 8 is also projected in the two-dimensional direction. Then, since they show the same tendency, it is concluded that the block corresponds to pattern 8.

(rV) Image region extraction Connections between each block may be examined on a block-by-block basis, or each block may be divided, for example, into two vertically and horizontally (total of 4
You can also examine it by dividing it into sections. FIG. 8 is a simplified explanatory diagram to show the latter method. FIG. FIG. 8(b) is a diagram in which the pattern numbers of FIG. 6 are attached to each block, and FIG. 8(c) is a diagram in which each block is divided into four as described above (for example, block α→small block α1°α2.α3. α4) and each obtained small block is given the pattern number in FIG. 6.

If each pattern in Figure 6 is divided into four, it is possible to know in advance which patterns each pattern will be decomposed into, so as shown in Figure 8 (
It is extremely easy to obtain Figure 8 (c) from (b).

In addition, if each pattern in Figure 6 is divided into four, it will be exactly
Each pattern in FIG. 6 may be selected so that it can be decomposed into the patterns shown in the figure, but it is sufficient that the patterns can be approximately decomposed into the patterns in FIG. Next, we examine the connections between small blocks, but in order to determine that there is a connection relationship,
For example, the judgment criterion may be based on the presence of at least a black part in both adjacent small block patterns near their joints, or the presence or absence of a connection may be judged by taking into consideration the overall shape of the patterns. good. In any case, the combinations of pattern connections are limited, so if you decide in what positional relationship each pattern should be adjacent to have a small block connection relationship, you can determine whether small blocks are connected or not. The decision is quite simple. Figure 8 (D) is a diagram showing the image area obtained by assuming that there is no connection relationship between small blocks when there is only a white part near the pattern junction. This is the image area of 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 compatibility, it is possible to reduce the size of the block from the beginning, but if this is done, the above-mentioned texture analysis will take a lot of time.

As a result of the above, the image area has been extracted.

Incidentally, the degree of appearance (frequency) of each pattern in the gradation drawing, the two character drawings, and the s; i drawing can be shown in FIG.

Therefore, by knowing the pattern distribution within each image area, it is also possible to determine what kind of image each image area is.

Figures 10 to 12 are more specific explanatory diagrams of area extraction, with Figure 10 showing the original image, Figure 11 showing the pattern classification image, and Figure 12 showing the area extraction image. It is a diagram. Note that the areas indicated by 1 and 2 in FIG. 10 are filled with characters, and the areas indicated by 3 and 4 are continuous tone images.

As shown in Fig. 12, image extraction is performed well, but in Fig. 11, isolated type 1
4 (block), which forms its own image area in FIG. In order to prevent this noise, an isolated block may be treated as being in the same image area as surrounding blocks.

(Effects of the Invention) As explained above, according to the method of the present invention, image regions of any shape can be extracted.

Therefore, great effects can be obtained when the method of the present invention is used when extracting only necessary parts and recording or filing them.

[Brief explanation of drawings]

FIG. 1 shows a procedure 70 in one example of the method of the present invention.
-T? - Fig. 2 is an explanatory diagram of the window, Fig. 3 is an explanatory diagram of the l-maro histogram, Fig. 4 is a flowchart showing the procedure for determining the background level, Fig. 5 is an explanatory diagram of dividing the image into blocks, and Fig. 6 is an explanatory diagram of the window. The figure is an explanatory diagram of the block pattern, No. 7
Figure 8 is an illustration of pattern classification, Figure 8 is an illustration of image area extraction, Figure 9 is an illustration of pattern frequency, Figure 10 is an original image, Figure 1, and Figure 11 is a diagram of a pattern classified image. , FIG. 12 is a diagram showing a region extracted image. α...Block α1. α2. α3. α4...Small []Tsuku1.2...
・Character drawing area 3.4...Continuous tone drawing area Patent applicant: Roku Konishi Photo Industry Co., Ltd. Agent: Fuji Ijima, Jigai: 1 name, Heron: 2 illustrations, 4 illustrations: 4 Ino 5 illustrations

Claims (5)

[Claims] (1) Divide an image into multiple blocks, classify each block into a predetermined pattern using binarized image data, and use the pattern to check connections between each block and extract similar image regions. Featured region extraction method. (2) The area extraction method according to claim 1, wherein the binarization threshold value when obtaining the binarized image data is determined based on background information of the image. (3) When classifying the blocks into patterns, data obtained by two-dimensionally projecting the binarized image data of each block is used. area extraction method. (4) When checking the connections between the blocks, each of the blocks is divided into two vertically and horizontally, and the connections between the divided small blocks and the surrounding small blocks are checked. 3. The area extraction method according to item 2, item 3, or item 3. (5) The area extraction method according to any one of claims 1 to 4, characterized in that an isolated block is regarded as the same image area as surrounding blocks.