JPS61175862A - Picture processing device - Google Patents

Abstract

PURPOSE:To enable proper space filtering processing according to picture quality of inputted picture by providing a means that extracts feature quantity of the picture quality from a variable density picture and a means that changes weight coefficient of a space filter according to the quantity of extraction. CONSTITUTION:A feature extracting circuit 8 extracts maximum value of picture element value from picture data stored in an input picture memory 1 as feature quantity relating to the picture quality. Then, one of two kinds of weight coefficients stored in a coefficient memory 10 is selected by a selecting table 9 according to the feature quantity. That is, out of input picture elements stored in the memory 1, picture elements centering around a remarked picture element are developed in registers 401-425, and products of each picture element value and weight coefficient of each element of a space filter selected by the selecting table 9 are obtained. Then, sum of the products is obtained by an adder 6 and each remarked picture element is stored in an output picture memory 7. Thus, proper space filtering processing can be made according to picture quality of an inputted picture.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image processing device used for spatial filtering processing, which is the most basic processing in figure recognition and the like.

Spatial filtering processes the i-th pixel in the right direction (X direction) and the j-th pixel in the downward direction (y direction) in image data expressed as multivalued data for each pixel arranged in a 1-te column. Assume that the pixel of interest is 1, for example, the pixel values of that pixel and its 8 neighboring pixels P(i-1, j-1) P(i, j-1) P(
i+l,j-1)P(i-1,j)P(i,
j) P (i+1.j)P(i-1,j+1)
P(i,j+1) P(i+1.j+1
), for example, the sum S (i, j) = -P (i-1, j-1) +
P (i+1.3-1)-P (i-1,j) +
P (i+1.j)-P (i-1,j+1) +
By finding P (i+1.j+1), find the density gradient (differential value) in the +X direction at that pixel, and process in the same way using the weighting coefficients of Figure 2 011), (C), +d). Accordingly, the concentration gradients in the -X direction, +y direction, and -y direction are determined.

At this time, it is desirable to be able to obtain good processing results for various images with different image qualities such as contrast.

[Conventional technology]

FIG. 3 is a block diagram showing the configuration of a conventional image processing device used for spatial filtering processing, and is similar to FIG. 1.

1 is an input 1ibJ image memory that stores a grayscale image as multivalued data of 0 to 255 expressed by 8 bits for each pixel arranged in a matrix of 512 rows x 512 columns.

2 is a coefficient memory that stores weighting coefficients of each element of a 5×5 pixel spatial filter.

1iI! 31 to 34 are grayscale images stored in the input image memory 1 that are read by rask scanning. A 512 pixel long shift register that sequentially shifts multi-value data for each l pixel.

401-405 are 1-pixel length registers (R) for sequentially shifting and temporarily storing multivalued data for each pixel obtained by reading the grayscale image stored in the input image memory l by rask scanning, 406-410.411 ~415°416~
420.421 to 425 are shift register 3, respectively.
Each pixel output from 1, 33, 33, and 34
11. Sequentially shifting and temporarily storing the multivalued data of
111 prime length register (R).

Adders 501 to 525 multiply each of the stored contents of the 25 registers 401 to 425 by each of the weighting coefficients of the 25 elements of the 5×5 spatial filter stored in the coefficient memory 2.

6 is 2 obtained by 25 multipliers 501 to 525.
An integrator that calculates the sum of five products.

7 is an output internal image memory that stores the output of the adder 6 for each pixel of interest.

In the above configuration, the registers 401 to 425 are 5
By configuring an arithmetic processing window of ×5 and reading the grayscale image stored in the input image memory l one by one by rask scanning with the main scanning direction in the X direction, the register 413
5×511 centering around the pixel value of the pixel of interest stored in
! ! The pixel values of l pixels are developed into registers 401 to 425, and nine multipliers 501 to 525 expand the pixel values to each register 401 to 425.
The product of the 425 pixel values and the weighting coefficient of each element of the 5×5 spatial filter stored in the coefficient memory 2 is obtained, the sum of these products is obtained by the adder 6, and the result is added for each pixel of interest. The image is stored in the output image memory 7.

[Problems to be Solved by the Invention J In the image processing device having the above configuration, 1. For example, when performing spatial differentiation of a gray scale image, the problem is uniformly differentiated whether it is a gray scale image with strong contrast or a gray scale image with weak contrast. Processing is performed using the same weighting coefficient.

By the way, the spatial differential value is originally better to be found in a 3x3 pixel area, but 1For example, when determining the direction of maximum density change in each pixel from the spatial differential value in two directions, the X direction and the When calculating the spatial differential in the pixel region No. 3, the direction of maximum density change cannot be obtained accurately, that is, in small steps, from a grayscale image with weak contrast.

Therefore, for grayscale images with weak contrast, it is necessary to obtain spatial differentiation in a wider pixel area of, for example, 5 x 5, but the problem with the conventional example described above is that such a request cannot be met. There is.

[Means for solving problems]

An image processing device according to the present invention includes an input image memory that stores a grayscale image as multivalued data for each pixel arranged in a matrix, and a feature extraction circuit that extracts feature quantities related to the image quality from the grayscale image. The above-mentioned problem is solved by including means for changing the weighting coefficient of the spatial filter according to the extracted feature amount.

(Function) In other words, it is possible to perform spatial filtering processing with different weighting coefficients depending on the image quality.This makes it possible to select, for example, the size of the pixel area for which spatial differentiation is to be obtained, and to adjust the image quality such as contrast. Good processing results can be obtained for various images with different values.

〔Example〕

The gist of the present invention will be specifically explained below using examples.

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention, and the same reference numerals as in the conventional example in FIG. 2 refer to the same objects. others.

8 is from the image data stored in the input image memory l,
A feature extraction circuit that extracts the maximum value of two pixel values as a feature related to the image quality.

9 and 10 are constituent elements of a means for changing the filtering coefficient of the spatial filter according to the feature quantity extracted by the feature extraction circuit 8, and 9 indicates that the maximum value detected by the feature extraction circuit 8 is 9, for example, 0. -127 or 128-255, as illustrated in FIG. 4, a selection table that stores the correspondence between spatial filters (a) and (bl) with different 1 weighting coefficients to be used. , 10 is a coefficient memory that stores weighting coefficients for each of the spatial filters as illustrated in FIG.

In the above configuration, first, the feature extraction circuit 8
For example, if the maximum value is 100, the maximum pixel value of the input image stored in the input image memory l is detected by the selection table 9. Among the two types of weighting coefficients, +8) is selected.

Thereafter, in the same way as in the conventional example, 5×5 pixels centered on the pixel of interest among the input image stored in the input image memory 1 are developed into registers 401 to 425, and selected based on each pixel value and the selection table 9. Figure 4 (Example 5×
The product of each element of the spatial filter of 5 with the weighting coefficient is obtained,
The sum of these products is obtained by the adder 6, and the result is stored in the output image memory 7 for each pixel of interest.

If the contrast of the input image is strong and, for example, the maximum value detected by the feature extraction circuit 8 is 200, the 5×5 pixel values developed in the registers 401 to 425 and the example shown in FIG. The product of each element of the 5×5 spatial filter with the weighting coefficient is obtained, and the sum of these products is obtained by the adder 6.

The results are stored in the output image memory 7 for each pixel of interest, and processing is performed using a spatial filter with a weighting coefficient of (a) or (bl) that differs depending on the contrast of the input image.

〔Effect of the invention〕

As explained above, according to the two aspects of the present invention, it is possible to perform appropriate spatial filtering processing according to the meat quality of the input image.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram of an embodiment of the present invention. FIGS. 2(a) to (dl) are explanatory diagrams related to fields of application. FIG. 3 is a block diagram of a conventional example. FIGS. 4+a and (b) are explanatory diagrams of an embodiment of the present invention. In the figure. 1 is an input image memory, and 31 to 32 are shift registers. 401 to 425 are registers, and 501 to 525 are multipliers. 6 is an adder, 7 is an output image memory. 8 is a feature extraction circuit, and 9 is a selection table. (αn
(C)C) (d) Mine 2fJ (α] Thing C#) I2I

Claims (2)

[Claims] (1) An image processing device that performs spatial filtering processing on a grayscale image, comprising an input image memory that stores the grayscale image as multivalued data for each pixel arranged in a matrix, and information on the image quality from the grayscale image. An image processing apparatus comprising: a feature extraction circuit that extracts a feature amount; and means for changing a weighting coefficient of the spatial filter according to the extracted feature amount. (2) The image processing apparatus according to claim (1), wherein the feature extraction circuit detects a maximum value of pixel values of the grayscale image.