JPH0351353B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION FIELD OF INDUSTRIAL APPLICATION The present invention relates to an image signal processing device having a function of reproducing a gradation image in binary.

Conventional configurations and their problems In recent years, the use of facsimiles in daily work has been expanding more and more, and along with this, there has been a growing demand for reproduction of gradation images in addition to the conventional black and white binary. In particular, binary pseudo gradation reproduction is strongly desired because it is well suited for display devices and storage devices.

Various dither methods are widely used for these pseudo gradation reproductions, in which the image is binarized according to a matrix table of threshold values. However, these conventional methods have a contradiction in that in order to improve tone reproducibility, it is necessary to make the matrix table large, and in order to obtain high resolution, the matrix table must be made small. It was difficult to achieve both high resolution. In particular, for images in which a gradation image and a binary image coexist, one of them has to be sacrificed.

The minimum average error method is a method that achieves both the above-mentioned gradation reproducibility and high resolution, and has been given a relatively good evaluation for reproduced image quality among various conventional methods.

FIG. 1 is a block diagram of the main parts of an apparatus for realizing the above-described minimum average error method.

In FIG. 1, 1 is an error data storage means, 2
3 is an error matrix, 3 is a storage device for the error E _xy at the coordinate xy of the original image, 4 is an average error calculation means that inputs the error data of the error matrix 2 and outputs the average error E〓, 5 is the coordinate xy of the original image The input terminal of the original image data I _xy in , 6 is I′ _xy (=I _xy +E _xy )
7 is an output terminal for a binary signal P _xy of 0 or R, 8 is a signal terminal to which a constant threshold value R/2 is applied, and 9 is a signal terminal that compares the input signal I' _xy with a constant threshold value R/2 to calculate I ' Binarization means outputs P _xy = R when _xy > R/2, and outputs P _xy = 0 in other cases; 10 indicates E _xy (=
I′ _xy −R _xy ).

Now, the calculation of the above average error E〓 will be explained as follows.

Eα=(1/ 〓 ^ij 〓 _ij ) 〓 ^ij 〓 _ij E _x+i,y+j ……(1) (However, i, j indicate the coordinates in error matrix 2.) Also, the coefficient matrix α _ij is It weights the error with respect to the distance from the _coordinates The operation of configuration 1 will be explained step by step. First, the average error calculation means 4 calculates and outputs the average error E from the contents of the error matrix 2,
Next, the calculation means 6 calculates the original image data I _xy at the input terminal 5.
and the average error E〓 are added and the resulting value I′ _xy is output. Then, the binarization means 9 compares the value I′ _xy with a constant threshold value R/2, and outputs a binary output P _xy to the output terminal 7. On the other hand, the calculation means 10 calculates the values I′ _xy and 2
A difference E _xy between the value output P _xy is calculated and outputted to the storage means 1 as a new error in the coordinate _xy of the original image.
Note that the error matrix 2 moves to the right in the figure on the storage means 1 in the main scanning direction of the original image every time one pixel of the original image data I _xy is input, and returns to the left end and moves down one line every sub-scanning. become.

Although the average error minimum method has better reproduced image quality than the simple dither method in which binarization is performed while referring to a matrix table of threshold values, there have been few examples of its application to general facsimile devices. The reason for this is that the amount of calculation required for image data processing is large and that a line buffer for storing error data is required, resulting in higher manufacturing costs than the simple dither method. In particular, when a line buffer is used to process a color image, the buffer capacity increases and the burden becomes heavier.

Purpose of the Invention In view of the conventional problems with the minimum average error method, an object of the present invention is to provide an image processing device that does not impair the gradation reproducibility of a binary image even if the storage capacity of the line buffer is reduced. .

Configuration of the Invention The present invention provides first input pixel data that has already been binarized.
an input means for sequentially inputting second and third input pixel data I _x,y and I _{x+1,y to be} binarized, which are consecutive pixels for I x-1 _, y; , has already stored the binarization error of the upper u bits caused by the binarization process of the first input pixel data I _x-1,y, and the second and third input pixel data I _{x . _ _} Second
moves along with the binarization process of the input pixel data I _x,y, and the neighboring position corresponding to the second input pixel data I _x,y is stored in the error data storage means.
a scanning window for reading out and outputting a predetermined number of binary errors of upper u bits; and a weighted average for calculating a weighted average value Ew from the output of the scanning window corresponding to the second input pixel data I _x,y. a calculation means; and a weighted average value calculated by the weighted average calculation means.
Ew is the upper bit, and the first processing pixel I _x-1,y
The average error with the lower bit data E _x-1,y [L] of the binarization error caused by the binarization process as the lower bit
Eα and the second input pixel data I _x,y , the first calculation means calculates the added value I′ _x,y =E〓+I _x,y ; and the first calculation means calculates. _The added value _{I '} converting means; and converting the second input pixel data I _{x,y from the added value I′ x,y} calculated by the first calculating means and the binary output P _{x,y that} is the output of the binarizing means. a second arithmetic means for calculating the binarization error E _xy = I′ _xy −P _xy ; and a binarization error E _xy calculated by the second arithmetic means.
At the same time, the upper u bits of the binary error _E
E _xy [L] is held in order to be sent to the first calculation means as the lower bit data of the average error Eα of the third input pixel I _x+1,y to be subjected to the next binarization process. The above object is achieved by providing a retaining means that includes a retaining means.

DESCRIPTION OF EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

FIG. 2 shows the blocks of an image signal processing device in one embodiment of the present invention.

In the figure, 1 to 3 are the same as the explanation in FIG. 1, but the number of data bits handled is reduced to U bits. 5 to 10 are the same as in FIG. Further, the weighted average calculation means 11 performs the same calculation as the average error calculation shown in FIG. 1 on U-bit data. The difference from the configuration shown in FIG. 1 is that a delay means 12 is provided, which delays the lower bit data of the binarization error by one pixel.

The operation of the configuration shown in FIG. 2 will be explained step by step below.

First, the weighted average calculating means 11 calculates the error matrix 2.
The calculation circuit 6 calculates and outputs the weighted average value E _w from the contents of , and uses the weighted average value E _w as the upper bit,
The average error Eα of the lower bit L bit data E _x-1,y (L) in the binarization error of the previous pixel, which is the output of the delay means 12, and the original image data at the input terminal 5
Add I _xy and output the value I′ _xy . The binarization means 9 compares the value I′ _xy with a constant threshold value R/2 and outputs a binary output P _xy to the output terminal 7. On the _{other hand, the calculation means 10 calculates the difference E xy between the value} I' Then, the lower (L) bit E _xy (L) is output to the delay means 12. The delay means 12 updates the previous pixel E _x-1,y (L) to new input data E _xy (L) and holds it. Note that the error matrix 2 is main-scanned on the storage means 1 to the right in the figure and sub-scanned downward.

When calculating the average error Eα, a weighted average is calculated by adding L bits of 0 to the lower part of the error data of each U beat in the error matrix, and adding it to the output data of the delay circuit to calculate the average. By setting the error to Eα, the calculation accuracy of the weighted average can be improved. Further, by adjusting the timing of data input/output in each block so that the previous calculation result is held in the calculation means 10 at the point where the calculation means 6 receives the input data Eα, the delay means 12 is unnecessary. becomes.

Experimental results show that reducing the lower 5 bits of 8-bit data does not cause any deterioration in image quality, and on the contrary, the visually unpleasant part of the striped texture characteristic of the conventional minimum mean error method is eliminated. I'm getting good results.

Effects of the Invention As described above, in the present invention, by correcting the data for L bits by the next pixel calculation, the gradation reproducibility is not impaired compared to the conventional minimum average error method, and the storage contents of the error data storage means are reduced to 1. L bits per pixel can be reduced.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a main part of an apparatus for implementing the conventional minimum average error method, and FIG. 2 is a block diagram of an image signal processing apparatus according to an embodiment of the present invention. 1... Error data storage means, 2... Error matrix, 3... Error data storage position of pixel of interest xy,
4... Average error calculation means, 5... Original image data input terminal, 6, 10... Addition/subtraction calculation means, 7... Binary signal output terminal, 11... Weighted average calculation means, 1
2... Delay means.

Claims (1)

[Claims] 1. Second and third input pixel data I to be binarized that are continuous as pixels with respect to the first input pixel data I _x-1,y that has already been binarized. _x,y ，I _x+1,y
input means for sequentially inputting the first input pixel data I _x-1,y in that order; 2. 2 of the third input pixel data I _x,y, I _x+1,y
error data storage means for sequentially storing binarization errors of upper u bits caused by the digitization process _;
moves along with the binarization process of the input pixel data I _x,y, and the neighboring position corresponding to the second input pixel data I _x,y is stored in the error data storage means.
a scanning window for reading out and outputting a predetermined number of binarization errors of the upper u bits; and a weight for calculating a weighted average value Ew from the output of the scanning window corresponding to the second input pixel data I _x,y. an average calculation means; and a weighted average value calculated by the weighted average calculation means.
Average error with Ew as the upper bit and lower bit data E _{x-1, y [L] of the binarization error caused by the binarization process of the first processing pixel I x-1,y} as the lower bit.
Eα and the second input pixel data I _x,y , a first calculation means for calculating the added value I′ _x,y, =Eα+I _x,y; Binarizing the added value I′ _x,y by comparing it with a threshold value, and making the result of the binarization the binary output P x _{,y of the second input pixel data I x ,y} the second input pixel data I _{x,y from} the added value I _′ a second arithmetic means for calculating the binarization error E _x,y = I′ _x,y −P _x,y ; and a binarization error E _x,y computed by the second arithmetic means.
The upper u bits of the binary error E _x,y are sent to the error data storage means, and the remaining lower bit data of the binary error E
In order to send E _x,y [L] to the first calculation means as lower bit data of the average error Eα of the third input pixel I _x+1,y to be subjected to the next binarization process. An image signal processing device comprising a holding means for holding the image signal. 2 The weighted average calculating means adds the lower L bits of "0" data to the binarization error of the predetermined number of upper U bits output by the scanning window, and then calculates the weighted average value.
An image signal device according to claim 1, which calculates Ew.