JPH0197650A - Image processing method - Google Patents

Abstract

PURPOSE:To remove a stripe pattern and a granular noise in a reproduced image, by a method wherein a weight proportion in dispersion of error data is made constant with respect to the same pixel. CONSTITUTION:When an error dispersion matrix is used, an error E1 generated in the case where an image data of a pixel of A12 is processed is dispersed to pixels of A13-A33. At that time, a product of a weighted coefficient Bij (i, j are natural numbers) and E1 is added to each pixel. For instance, an error to be added to the pixel A22 is expressed by E1XB22. Then, when E2 is taken as an error in execution of binary processing of A13, E2 is also dispersed to A14-A34 in the same way as abovementioned, and an error of E2XB22 is dispersed to A22. At this time, when the coefficient B22 is made larger than the other coefficients, errors can be much dispersed to the pixel of A22, dots are prevented from dispersion in a part where density of a copy is uniform, and granular noise can be decreased. Further, a stripe pattern can be also prevented in a shadow part of the copy.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image processing method in digital printers, digital facsimiles, etc., and particularly relates to an image processing method that satisfactorily reproduces halftone images.

[Conventional technology]

2. Description of the Related Art Error diffusion has conventionally been used as a binarization method for reproducing halftones in digital printers, digital facsimiles, and the like. This method calculates the density difference for each pixel between the image density of the original and the output image density, and distributes the error that is the result of this calculation after applying a specific weight to the surrounding images (there is a method). Regarding the diffusion method, see Reference R,
W.

Floyd and L, Steinberg”A
n Adaptive Algorithm for
5peacial Gray 5cale”
(1976).

[Problem that the invention is trying to solve]

In the binarization processing power formula using the error diffusion method described above, the moiré phenomenon that is a problem with other binarization methods such as dithering and density patterning does not occur, so that text and photographs can be reproduced well. However, unique striped patterns occur in uniform density areas of the image (other than edge areas), and grainy noise becomes noticeable especially in the highlight areas of the image due to scattered dots, resulting in a decrease in image quality. There was a drawback that

[Means and operations for solving the problems] The present invention aims to solve the above-mentioned problems by distributing error data generated during the binarization process of an image to surrounding pixels, and converting the halftone image into a halftone image. In an image processing method that performs reproduction, the weighting ratio when distributing error data is kept constant for the same pixel.

In other words, according to the method of the present invention, the striped pattern and granular noise in the reproduced image are removed by making the ratio of errors from a plurality of pixels to one pixel constant.

〔Example〕

Embodiments of the present invention will be described in detail below with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention.

The input sensor section 1 includes a photoelectric conversion element such as a CCD and a driving device for scanning the element, and reads and scans a document. The image data of the document read by the input sensor section 1 is sent to the A/D converter 2. Here, the data of each pixel is converted into digital data of 8t>it, and quantized into data having 256 levels of gradation.

Next, in the correction circuit 3, shading correction and the like for correcting sensitivity unevenness of the CCD sensor and illuminance unevenness due to the illumination light source are performed by digital calculation processing.

The data corrected by the correction circuit 3 is sequentially stored in the image memory 4. The image memory 4 is a multi-level memory that can store data for at least one document.

The binarization circuit 5 performs binarization processing using the error diffusion method. The error diffusion weighting to be performed is constituted by section 5-2. Next is Figure 2.

The binarization process in this embodiment will be explained using FIG. 3.

FIG. 2 shows one part (8 pixels) of image data, and Aij (i and j are natural numbers) each represent one pixel.

First, the error E1 generated when processing the image data of the pixel Als is A Il+ A! when using the error diffusion matrix shown in FIG. 3(a). ++ A 211 A! I+
Each of the pixels is diffused into A II+ A 11 A as pixels.

At this time, the product of the weighting coefficient Btj (t, j are natural numbers) and El is added to each pixel in the weighting shown in FIG. 3(a). For example, the error added to pixel A i2 in FIG. 2(a) is E
It can be expressed as 1xB2z.

Next, after the binarization process of AH is completed, the error is calculated by shifting one pixel and performing the binarization process of A shown in Fig. 2(b).
Set it to 2. This E2 is also AI4. A2*
, Is, Aha, A1*, Ass,
Spread to A34. A again here. Looking at the error diffused into A, 2g and B, as mentioned above. The error diffusion matrix in FIG. 3(a) is shifted as shown in FIG. 3(b) so that A x2
The error of E 2 x B I2 is diffused.

At this time, if the weighting coefficient B 12 is set to a larger value than the other weighting coefficients, it is possible to always diffuse more errors to the A 22 pixel, which makes it possible to perform pseudo halftone dot formation. This prevents the dots from dispersing in areas where the density of the document is negative, especially in the highlight areas, and the dots are arranged periodically, making it possible to reduce grainy noise felt in the highlight areas. In other words, since the density of the image is low in the highlight portion, 0 is often output when binarized at a constant slice level. In reality, images have density, so the error (+E) between O after this binarization process and the image density is
) is diffused to surrounding pixels.

Therefore, if a certain pixel always has a lot of undertones, the dots can be arranged periodically so that the dots are turned on in that part. Noise can be prevented.

4 and 5 show the error diffusion processing section 5-1 of FIG. The error diffusion weighting is a diagram showing details of the section 5-2.

The binarization process using the error diffusion method in this embodiment will be described below with reference to FIGS. 4 and 5.

20 is an error calculation circuit; 28 is a comparator that binarizes input data using a threshold value T (for example, 127) and outputs an output Z (O or 1); 36 to 42 are image data and error E output from the error calculation circuit; A D flip-flop (hereinafter referred to as DFF) that stores the sum of the image data and the diffusion error (diffusion error), 21 to 27 are adders that calculate the sum of the image data and the diffusion error, and 28 to 35 are the error calculation circuits 20. A multiplier that calculates the product of the error E outputted from the weighting coefficient (Bij), 43 and 44 a line memory that stores image data for pixels excluding 3 dots from one line of image data; is the image memory in FIG.

FIG. 5 shows a circuit for shifting the weighting coefficient in response to the pixel shift in FIG. 4 when the error E is diffused to each pixel.

45 to 53 are weighted DFFs that shift the coefficient by one, 5
Line memories 4 to 56 store weighting coefficients excluding three weighting coefficients from one line.

The operation of the circuits shown in FIGS. 4 and 5 will be explained below.

Image data is sequentially output from the image memory 4.

Here, the line buffer 43 stores image data of a pixel on the same line as the pixel of interest to be binarized by the comparator, and the line buffer 44 stores image data of a pixel one line after that to be processed thereafter. There is.

First, image data to which error data from pixels for which the binarization process has already been completed is output from the DFF 36 and binarized by the comparator 28 using a threshold value 127. At this time, for example, if the output from the DFF 36 is 30, 0 (white) is output as the binary output Z. In the error calculation circuit 20, D
The error +30 between the output 30 from the FF 36 and the binary output O is output as error data E. This 30 is then diffused to unprocessed surrounding pixels. Figure 5 shows the weighting coefficients used during this diffusion, and the actual Bij
shows the value of

The weighting in FIG. 6 is configured such that the coefficients are weighted at a higher rate at the center of the block (9 pixels) than at other pixels.

The error data E outputted from the error calculation circuit 20 is multiplied by a weighting coefficient in multipliers 28 to 35, and added to the image data in adders 21 to 27, respectively. The added image data is added to DFFs 36-42. This completes the processing of one pixel.

Upon completion of the process, a clock (CLK) is input from a control section (not shown) to the circuits shown in FIGS. 4 and 5, and both the pixel data and weighting coefficients are shifted by one. Thereafter, binarization processing is performed in the same manner. By repeating the above processing, the binarization processing using the error diffusion method in this embodiment is performed.

The error diffusion matrix and weighting coefficients are shown in Figure 3.
It is not limited to FIG.

In this way, according to the above-described embodiment, it is possible to concentrate the error diffusion by the error diffusion method on a predetermined pixel, and to form a pseudo halftone dot.

This prevents dots from dispersing in areas where the density of the document is uneven, particularly in the highlight areas, and since the dots are arranged periodically, it is possible to reduce grainy noise felt in the highlight areas.

Further, even in the shadow portion of the document, by keeping the error weighting ratio for each pixel constant, it is possible to prevent a unique striped pattern caused by the shape of the error diffusion matrix.

〔Effect of the invention〕

As explained above, according to the present invention, two
Since the rate of diffusion of errors that occur during the value conversion process is made constant for the same pixel, it is possible to prevent striped patterns and granular noise that occur during the error diffusion method.

[Brief explanation of the drawing]

Fig. 1 is a block diagram showing an embodiment of the present invention, Fig. 2 is a diagram showing error diffusion, Fig. 3 is a diagram showing an error diffusion matrix, and Figs. 2 [Details of conversion circuit,
FIG. 6 is a detailed circuit diagram showing an example of weighting coefficients during error diffusion. In the figure, 1 is an input sensor section, 2 is an A/D converter, 3 is a correction circuit, 4 is an image memory, 5 is a binarization circuit, and 6 is a printer.

Claims (1)

[Claims] In an image processing method that reproduces a halftone image by distributing error data generated during image binarization processing to surrounding pixels, the weighting ratio when dispersing error data is kept constant for the same pixel. An image processing method characterized by: