JPH0376629B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an image processing method that performs at least one of enlarging or reducing an image and filtering (image enhancement, etc.).

(Prior Art) As methods for enlarging or reducing a binarized image using a pixel density conversion method, the SPC method, the logical sum method, the nine-division method, the projection method, and the like have been known.

(Problems to be Solved by the Invention) Problems with these conventional methods include the fact that the lines are conspicuously blurred or missing.
The biggest problem is that moiré fringes occur when enlarging or reducing a halftone image with a periodic structure, such as a binarized image expressed by the systematic dither method.

On the other hand, in the case of this type of binarized halftone image, there is no method to easily perform filtering.

The present invention has been made in view of this point, and its purpose is to perform image processing that can enlarge or reduce even a binarized image with a periodic structure without causing moiré fringes, and also allows easy filtering. The purpose is to provide a method.

(Means for Solving the Problems) The present invention, which solves the above-mentioned problems, reads out a binarized image stored in a memory, and reads out a binarized image stored in a memory, A first step of dividing the image into blocks, counting at least one of the number of black pixels or the number of white pixels for each block, and storing the counted value in memory; and storing the number of pixels obtained in the first step from the memory. A second device that reads out the number of pixels, performs filtering on the number of pixels, and stores the number of pixels after filtering in the memory.
The number of pixels after filtering obtained in the step and the second step is read from the memory, and from the number of pixels, a first density matrix pattern corresponding to each number of pixels is obtained for each block. The third step is to store the first density matrix pattern in the memory, and the first density matrix pattern of each block obtained in the third step is read out from the memory, and is calculated according to the magnification m/n (where m and n are integers). When converting to the second density matrix pattern of the size, the density value of the x row and y column in the second density matrix pattern of the block located Ith in the row direction and Jth in the column direction is In the first density matrix pattern of the block before enlargement and reduction located at the I-th and J-th column,
Adopt the density value of AD1 row and AD2 column.
mod[p, q] is the remainder of p÷q.

For magnification (magnification m/n>1), AD1=mod[x+mod[(I-1)(m-n),n]+1-2,n]+1 AD2=mod[y+mod[(J-1)( m-n) ,n]+1-2,n]+1 For reduction (magnification m/n<1), AD1=mod[x+mod[(I-1)(n-|n-m
|) ,n]+1-2,n]+1 AD2=mod[y+mod[(J-1)(n-|n-m
|) ,n]+1-2,n]+1) A fourth step of obtaining the second density matrix pattern in all blocks, and reconstructed with the second density matrix pattern of each step. The feature is that the image is a processed image.

(Example) Hereinafter, the image processing method of the present invention will be specifically explained using FIGS. 1 to 7.

First, the object of image processing is a binarized image. This binarized image can be easily obtained by using, for example, a 4×4 or 8×8 dither matrix as a threshold value. The setting width of the threshold values that make up this dither matrix is, for example, wide in the case of gradation drawings, about 0.1 to 1.4 in reflection density, and narrow in the case of line drawings, about 0.1 to 0.5 in reflection density (or by setting a fixed threshold value). It is preferable to use This is to prevent the image from becoming blank or thick. Further, different types of dither matrices may be used for gradation drawings and line drawings. Note that the binarization method may be a method other than the dither method, such as a density pattern method or a dot method.

In the method of the present invention illustrated in FIG. 1, first, if the image is not a binarized image, the above-mentioned binarization is performed (step). In Figure 2, the size is 4×
Using the dot dispersion type (Bayer type) dither matrix DM1 (see Fig. 2 A) of No. 4, the original image A (see Fig. 2 B) is binarized by the dither method, as shown in Fig. 2 C. An example of obtaining a binarized image B is shown. In this figure, the dither matrix
The numbers in DM1 and original image A indicate standardized density levels, and the pixels in the diagonally shaded portion of binarized image B indicate that they are black pixels.

In the next step, the binarized image is divided into blocks of appropriate size. In FIG. 2C, it is divided into 4×4 sizes. Then, the number of black pixels (or the number of white pixels) in each block is counted (step), and the process moves to filtering. The filtering performed here converts the number of black pixels (hereinafter referred to as the original number of black pixels) to another number of black pixels (hereinafter referred to as the converted number of black pixels) using a predetermined spatial filter. This is done by doing (steps). In the example of FIG. 3, the spatial filter shown in FIG. 3A is used to convert the original black pixel number shown in FIG. 3B to the converted black pixel number shown in FIG. 3C. Note that in order to filter the outermost block, data on the number of black pixels on the outer side is also required, so for the sake of explanation here, the numbers shown in Braille are given as virtual data to perform filtering. What kind of filtering is done is
Determined by the spatial filter used during the transformation.
For example, image enhancement can be performed using the spatial filters shown in FIG. 3A and FIG. 4. Here, α in FIG. 4 is a natural number of 20 or less, and β is a constant. If α is large, the edge enhancement will be quite strong.

From the number of converted black pixels thus obtained, the density level of each block is determined to obtain a first density matrix pattern (step). Here, the block size is equivalent to the size (4 x 4 or 8 x 8) of the dither matrix (threshold value group) used to obtain the binarized image, preferably larger than the size of the dither matrix. Choose a smaller size. In this way, high resolution can be maintained while increasing the number of gradations. Figure 5(b) shows the number of converted black pixels in each block as the normalized average density level of each block, and Figure 5(c) shows the number of converted black pixels in each block as the normalized average density level of each block. In this example, the first density pattern is determined by copying the dither matrix DM2 (see Fig. 5A), which is the same as the dither matrix DM1 described above, and each block. This is done by comparing the concentration level of For example, in the case of block BK1, its density level is 9.
Therefore, the dither matrix DM in Figure 5 A
2, the portion with a density level of 9 or less becomes a black pixel, forming a density matrix pattern like the block BK1 in FIG. 5C.

After obtaining this first density matrix pattern, a second density matrix pattern of a size corresponding to the enlargement/reduction magnification is obtained for each block (step), and these are arranged in block order to form enlarged/reduced images. Get (step). Figure 6A is an enlarged image obtained in this way with a magnification of 5/4.
Figure b is a reduced image with a reduction magnification of 3/4. here,
The size ratio of the first and second density matrix patterns corresponds to the vertical and horizontal enlargement and reduction magnification,
The vertical and horizontal sizes of the second density matrix pattern are the vertical and horizontal sizes of the first density matrix pattern multiplied by the vertical and horizontal expansion/reduction magnification. Therefore, in the example of FIG. 6, the size is 5×5 in the case of FIG. 6A, and the size is 3×3 in FIG. 6B.

In this example, when the mother pattern obtained by two-dimensionally arranging the corresponding patterns in the first density matrix pattern is divided by the size of the second density matrix pattern,
Cut out a pattern that has the same positional relationship as the second density matrix pattern that you are trying to find,
This is used as the corresponding second density matrix pattern. FIG. 7 is an explanatory diagram (magnification: 5/4) showing an example of obtaining a second density matrix pattern from a first density matrix pattern.
4) Blocks f ₁₁ , f ₁₂ , f ₁₃ ,...f ₂₁ , f ₂₂ ,...,...
The plane consisting of the blocks F ₁₁ , F ₁₂ , F ₁₃ , of the second density matrix pattern (size 5×5)
..., F ₂₁ , F ₂₂ , ..., ... is shown. First, as a second density matrix pattern corresponding to block F ₁₁ , all blocks f ₁₁ ,
The position of block F ₁₁ is determined from the mother pattern assuming that f ₁₂ , f ₁₃ , ..., f ₂₁ , f ₂₂ , ..., ... contain the same first density matrix pattern corresponding to block f ₁₁ . Use a pattern cut out from . Similarly, the second density matrix pattern corresponding to block _F12 includes all blocks.
Blocks are added to f ₁₁ , f ₁₂ , f ₁₃ ,…, f ₂₁ , f ₂₂ ,…,…
From the mother pattern, which is assumed to contain the same first density matrix pattern corresponding to f ₁₂ ,
Use the pattern cut out at block F ₁₂ . That is, as a second density matrix pattern corresponding to block Fij, the same first density matrix pattern corresponding to block fij is applied to all blocks f ₁₁ , f ₁₂ , f ₁₃ , ..., f ₂₁ , f ₂₂ , ..., ... A pattern at the position of block Fij is cut out from a mother pattern that is assumed to contain a density matrix pattern. The above-mentioned FIG. 6A shows the second density matrix pattern obtained in this way arranged on a plane. The same applies to Figure 6B.

Note that matrices DM1 and DM2 do not need to be composed of the same thing; for example, if matrix DM1 and DM2 are
2 may be of dot concentration type (spiral type).

Further, a specific method for obtaining the pattern of each block after scaling described above may be to actually configure the above-mentioned mother pattern in memory and read data at a predetermined address from it to obtain the pattern. , this would require an extremely large memory capacity. Therefore, instead of actually creating a mother pattern, attention may be paid to the periodicity of the pattern, and the density of each pixel may be determined in the following manner to obtain a pattern after scaling.

That is, in the case of enlargement (magnification m/n), the x-th pattern in the I-th block in the row direction of the new block (block after enlargement) is the same as the pattern in the I-th block in the row direction of the block before enlargement. AD1=mod[x+mod[(I-1)(m-n),n]+1-2,n]+1th row pattern and y column in the Jth block in the column direction of the new block. The eye pattern is equal to the pattern of the AD2=mod[y+mod[(J-1)(m-n),n]+1-2,n]+1st column in the Jth block in the column direction of the block before expansion. , on the other hand, in the case of a small size (magnification m/n), the x-th pattern in the I-th block in the row direction of the new block (block after reduction) is the same as the pattern in the I-th block in the row direction before reduction. , AD1=mod[x+mod[(I-1)(n-|n-n
|) ,n]+1-2,n]+1 The pattern in the y-th column in the J-th block in the column direction and which is equal to the pattern in the J-th row in the column direction of the new block is the same as the pattern in the J-th block in the column direction of the block before reduction. , AD2=mod[y+mod[(J-1)(n-|n-m
|) ,n]+1-2,n]+1 Equivalent to the pattern in the 1st column.

Here, mod [p, q] is the remainder of p÷q,
Of course it is smaller than q.

Therefore, if the pixel density of the block before scaling is stored in memory, the new pixel density can be easily read by using AD1 and AD2 as addresses in the row and column directions and reading out the pixel density in the block before scaling. You can get the block pattern.

FIG. 8 is a diagram showing an example of an apparatus for realizing the method of the present invention. In the figure, 1 reads the binarized image stored in the memory 2, divides it into blocks whose size is less than the threshold value group used to obtain the binarized image, and counts the number of black pixels for each block. and a black pixel counting means 2 for storing the counted value in the memory 3, which reads the number of black pixels from the memory 3 and performs filtering using a desired spatial filter;
A filtering means 6 stores the number of black pixels after filtering in the memory 5, and 6 reads the number of black pixels after filtering from the memory 5, and calculates the number of black pixels for each block from the number of black pixels. A first density matrix pattern creating means 8 reads out the first density matrix pattern of each block from the memory 7 and stores the first density matrix pattern corresponding to the first density matrix pattern in the memory 7.
This is second density matrix pattern creation means for converting into a second density matrix pattern of a size corresponding to the magnification. In this second density matrix pattern creation means 8, when converting the first density matrix pattern into a second density matrix pattern, the second density matrix pattern of the block located I-th in the row direction and J-th in the column direction is converted. The density value of the x-th row and y-th column in the density matrix pattern is the aforementioned AD1 row in the first density matrix pattern of the block before enlargement/reduction located in the I-th row direction and J-th column direction. , adopt the concentration value in the second column of AD. Then, the second density matrix pattern creating means 8 stores the second density matrix patterns in all blocks in the memory 9. In this way, a binarized image reconstructed using the second density matrix pattern of each block, that is, a binarized image that has been enlarged or reduced and filtered, is stored in the memory 9. become. Incidentally, if the plurality of memories mentioned above have a large memory capacity, a single memory is sufficient.

(Effects of the Invention) As explained above, in the present invention, a second density matrix pattern having a size corresponding to the magnification is obtained for each block from the first density matrix pattern. is unchanged;
Unlike the conventional method of obtaining enlarged or reduced images by changing the sampling interval, moiré fringes are less likely to occur. Furthermore, since image enhancement is achieved by applying filtering to the average density level (number of black pixels) of a block, image enhancement and the like can be easily performed.

[Brief explanation of drawings]

FIG. 1 is a flowchart showing an example of the method of the present invention, FIG. 2 is an explanatory diagram showing an example of binarization in FIG. 1, FIG. 3 is an explanatory diagram of filtering in FIG. 1, and FIG. Explanatory diagram of spatial filter, 5th
The figure is an explanatory diagram of the first density matrix pattern in Fig. 1, Fig. 6 is an explanatory diagram of the second density matrix pattern (enlarged and reduced image) in Fig. 1, and Fig. 7 is an explanatory diagram of the first density matrix pattern in Fig. 1. An explanatory diagram of a method for obtaining a second density matrix pattern from a matrix pattern, FIG. 8 is a diagram showing an example of an apparatus for realizing the method of the present invention. DM1, DM2...Dither matrix, A...
...Original image, B...Binarized image, f ₁₁ , f ₁₂ ,
..., f ₂₁ , f ₂₂ , ... blocks of the first density matrix pattern, F ₁₁ , F ₁₂ , ..., F ₂₁ , F ₂₂ , ...
Block of second density matrix pattern.

Claims (1)

[Claims] 1. Read out a binarized image stored in a memory, divide it into blocks whose size is less than or equal to the threshold value group used to obtain the binarized image, and calculate the number of black pixels or the number of black pixels for each block. A first step of counting at least one of the number of white pixels and storing the counted value in a memory, and reading out the number of pixels obtained in the first step from the memory, applying filtering to the number of pixels, and filtering the number of pixels. The second one stores the number of pixels after ringing in memory.
step and the number of pixels after filtering obtained in the second step is read from the memory, and from the number of pixels, a first density matrix pattern corresponding to each number of pixels is obtained for each block. a third step in which the first density matrix pattern of each block obtained in the third step is read out from the memory and is processed according to the magnification m/n (where m and n are integers); When converting to the second density matrix pattern of the same size, the row In the first density matrix pattern of the block before enlargement or reduction located at the I-th direction and the J-th column direction.
Adopt the density value of AD1 row and AD2 column.
mod[p, q] is the remainder of p÷q. For magnification (magnification m/n>1), AD1=mod[x+mod[(I-1)(m-n),n]+1-2,n]+1 AD2=mod[y+mod[(J-1)( m-n),n]+1-2,n]+1 For reduction (magnification m/n<1), AD1=mod[x+mod[I-1)(n-|n-m
|) ,n]+1-2,n]+1 AD2=mod[y+mod[(J-1)(n-|n-m
|) ,n]+1-2,n]+1) A fourth step of obtaining the second density matrix pattern in all blocks, and reconstructing the second density matrix pattern of each block. An image processing method characterized in that an image is a processed image.