JPH06164935A - Picture processing method - Google Patents

Abstract

PURPOSE:To reproduce a picture excellent in the reproducibility of gradation and resolution by selecting a matrix having at least one difference from a size, a pattern and a weighting coefficient of matrices having the weighting coefficient used to share an error of picture data caused by binarization processing for each picture element so as to spread the error. CONSTITUTION:A comparator 1 compares picture data 5 with a threshold level 4 to quantize a picture signal and binarized data outputted from the comparator 1 are given to a subtractor 2, in which a difference from the input picture data is obtained and an error is calculated. A matrix selector 7 selects any of matrices 8a-8e and the calculated error is spread by a coefficient device 6 by using the selected matrix and the result is added to picture data 5 before binarization. The weighting coefficient, matrix size and distribution in the weighting coefficient differ from the matrices 8a-8e and any of the matrices 8a-8e is selected by the matrix selection section 7 for each picture element.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method for quantizing image data, and more particularly to an image processing method for quantizing image data using an error diffusion method.

[0002]

2. Description of the Related Art Conventionally, an error diffusion method has been used as an image processing method for reproducing a halftone image in a digital copying machine, a facsimile or the like. In this method, the difference (error amount) between the read input image data and the output data is calculated, and the error data as the calculation result is weighted and dispersed in the peripheral pixels before the binarization process. Is.

This error diffusion method will be described with reference to FIGS. In a digital copying machine, a facsimile, etc., as shown in FIG. 2, an original image is read by an image reading unit 10 including a line sensor, shading correction,
An image processing unit 11 that performs density conversion, binarization processing, and the like performs image processing, and an image output unit 12 including a printer outputs image data.

The error diffusion method is carried out in the binarization processing in the image processing section 11, and as shown in FIG. 3, the input image data X to the error correction circuit 20 and the binarization processing circuit 21 are binarized. The error calculation circuit 22 calculates the error component E from the binarized image data P, and the error component E is added to or subtracted from the input image data X to obtain a corrected image Z and binarized. is there.

In the binarization process, as shown in FIG. 4, when the threshold level is TH, the white level is 0, and the black level is 255, the image data X1 is larger than the threshold level TH. 255 ", image data X2
Is smaller than the threshold level TH and is therefore set to "0". In this binarization process, an error of d1 occurs between the image data X1 and the level 255, and an error of d2 occurs between the image data X2.
Therefore, this error is distributed to the peripheral pixels before the binarization process with a predetermined weighting, and the density is stored.

As a method of distributing the error, for example, as shown in FIG. 5 (a), assuming that the pixel of interest is X at present, the error is distributed around it by 4, 1, 4, 1. For example, for the image data X1, since data smaller than the level 255 is set to the level 255, d1 × 4/10 is subtracted when the coefficient is 4 and d1 × 1 / is calculated when the coefficient is 1.
Diffusion by subtracting 10.

Since the level d2 of the image data X2 is set to level 0, d2 × 4/10 is added when the weighting coefficient is 4 and d2 is added when the weighting coefficient is 1.
Disperse x1 / 10.

The weighting coefficient is, for example, as shown in FIG.
A matrix pattern such as 2, 1, 1, 2, 1 as shown in FIG. 4 may be used, and the coefficient, the pattern, and the size of the matrix pattern for the error diffusion can be appropriately changed.

In this error diffusion method, the error data between the input data density and the output image density is spatially adjusted, so that
Unlike the dither processing which is another image processing method, there is no restriction on the number of gradations depending on the matrix size, the output image density faithful to the original density can be reproduced, and both gradation and resolution can be compatible.

[0010]

By the way, in the above error diffusion method, as shown in FIGS. 5 (a) and 5 (b), it is common to increase the weighting coefficient of the pixel adjacent to the pixel of interest. As shown by an arrow A in FIG. 5A and an arrow B in FIG. 5B, the weighting coefficient has a directional property, which causes the density to have regularity, that is, a texture (striped pattern). there were.

In order to overcome this point, various improvement methods have been proposed, for example, a method of combining an error diffusion method with a dither matrix has been proposed, but there is a problem that the processing becomes complicated.

An object of the present invention is to solve the above problems, and an object of the present invention is to provide an image processing method capable of reproducing a high quality image by removing the texture generated by the error diffusion method with a simple structure. And

[0013]

SUMMARY OF THE INVENTION According to the present invention, an image signal captured from an image reading device is processed by using an error diffusion method.
One of a plurality of matrices in which at least one of a size, a pattern, and a weighting coefficient having a weighting coefficient for allocating an error amount of image data generated by the binarization processing is made different in the binarizing image processing method. Is selected for each pixel to perform image processing.

[0014]

The present invention prepares a plurality of matrices in which the size of the matrix for diffusing the error, the weighting coefficient, or the distribution method of the weighting coefficient is changed, and the matrix is switched for each pixel to input the input image. By distributing the error between the data and the output image data to the peripheral pixels, it is possible to easily prevent the occurrence of texture and reproduce a high-quality image.

[0015]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a diagram showing a configuration for performing error diffusion according to the present invention. In the figure, 1 is a comparator, 2 is a subtractor, 3 is an adder, 4 is a threshold value, 5 is image data, 6 is a coefficient unit, 7 is a matrix switching unit, and 8a to 8e are error distribution matrices.

In the figure, a comparator 1 is for comparing the image data 5 with a threshold value 4 to quantize the image signal,
The difference between the binarized data output from the comparator 1 and the input image data is calculated by the subtractor 2, and the error is calculated.
The calculated error amount is selected by the matrix switching unit 7 in any of the matrices 8a to 8e, dispersed by the coefficient unit 6 using the selected matrix, and this is added to the image data 5 before binarization ( Or subtract). The matrices 8a to 8e have different weighting coefficients, matrix sizes, or distribution methods of weighting coefficients, and these are switched by the matrix switching unit 7 for each pixel and used.

As a result, in the conventional constant matrix method, there is a problem that texture is likely to occur due to the periodicity of the weighting coefficient.
By changing at least one of the matrix pattern and its weighting coefficient for each pixel, it is possible to prevent the occurrence of texture.

[0018]

As described above, according to the present invention, at least one of the size, the pattern, and the weighting coefficient of the matrix having the weighting coefficient for distributing the error amount of the image data generated by the binarization processing is made different. By selecting one of a plurality of matrices for each pixel and diffusing the error, the texture that has been generated by the error diffusion method in the past can be removed, so that the original density can be faithfully reproduced, the gradation and the resolution can be improved. It is possible to reproduce a high-quality image with excellent reproducibility.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration for performing error diffusion of the present invention.

FIG. 2 is a diagram showing a configuration of an image processing apparatus.

FIG. 3 is a diagram illustrating an error diffusion circuit.

FIG. 4 is a diagram illustrating a binarization process.

FIG. 5 is a diagram illustrating generation of texture by an error diffusion method.

[Explanation of symbols]

1 ... Comparator, 2 ... Subtractor, 3 ... Adder, 4 ... Threshold value, 5 ...
Image data, 6 ... Coefficient unit, 7 ... Matrix switching unit, 8
a to 8e ... Matrix for error dispersion.

Claims (1)

[Claims] 1. In an image processing method for binarizing an image signal taken in from an image reading device by using an error diffusion method, a matrix having a weighting coefficient for allocating an error component of image data generated by the binarizing process. An image processing method characterized by selecting one of a plurality of matrices in which at least one of the size, pattern, and weighting coefficient is different for each pixel and performing image processing.