JP2001111831A - Image processing method - Google Patents

Abstract

(57) [Problem] To provide an image processing method for suppressing the occurrence of white spots in a black background and the occurrence of black spots in a white background and improving the contrast of a halftone image. SOLUTION: A result of diffusing an error between input pixel data and output pixel data quantized for each pixel of multi-value pixel data inputted for each line into peripheral pixel data is based on a predetermined threshold value. And binarize and output 2
In an image processing method for outputting value pixel data, an all black / all white determination unit 32 is provided for determining whether all the pixels of input multi-value pixel data are all black or all white, and all black or all white is provided for each pixel. An image processing method for setting an error from input pixel data to "0" when pixel data determined as all black or all white continues for a predetermined number of times.

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, and more particularly to an image processing method for processing a halftone image by an error diffusion method.

[0002]

2. Description of the Related Art Recently, in an information-oriented society, devices handling image data are increasing. In an apparatus that handles such image data, an image processing apparatus that converts pixel data read by a scanner or the like into a binary image signal is required. Recently, the image processing apparatus handles not only monochrome binary images such as characters but also halftone images such as photographs. Therefore, conventionally, as a processing method of a halftone image, a level of each pixel of multi-valued pixel data is compared with matrix data in which a plurality of predetermined threshold values are set, and binarization processing is performed based on the comparison result. A dither method for outputting binary pixel data to represent a halftone image, and converting a binary error between output pixel data and input pixel data quantized for each pixel of multi-value pixel data into peripheral pixel data. There is an error diffusion method in which a result of the diffusion is binarized based on a predetermined set threshold value, and output binary pixel data is output to express a halftone image. Attention has been focused on an error diffusion method that achieves a balance between resolution and resolution.

[0003]

However, in such a conventional error diffusion system, when the input pixel data is at all black levels or all white levels, no binarization error occurs, If the accumulated error of the pixel data up to that time is stored and the threshold value for binarization is always constant, a texture peculiar to error diffusion (an image tailing phenomenon that occurs when error diffusion is not sufficient) occurs. To eliminate this, there is a method of periodically changing the threshold value of binarization. However, a white point is generated near a black background, a black point is easily generated near a white background, and an output image is easily blurred. There was a problem.

[0004] In addition, for example, Japanese Patent Laid-Open Publication No.
In the error diffusion process executed by the image processing apparatus described in Japanese Patent Application Publication No. 72-72, the remainder (after the decimal point) of an error generated when a binarization error is obtained by a weighting operation.
Is added to the next pixel data to be correlated, corrected, the corrected pixel data is binarized by a certain threshold value, and the process of storing the input image density and the output image density and reproducing the image is performed. However, this apparatus cannot remove the above-mentioned texture peculiar to error diffusion.

Therefore, according to the present invention, when error diffusion processing is performed, when it is determined whether all pixels are all black or all white, the binarization error of all black and all white data is set to "0". Regardless of the error between the input pixel data of black and all white data, all black data or all white data is used, and the threshold value is set to the all white level for all black data and to the all black level for all white data. By binarizing the pixel data set to the level with a threshold value that is periodically changed for each processing pixel, the occurrence of white points in a black background and the occurrence of black points in a white background are suppressed, and the halftone image An object of the present invention is to provide an image processing method for improving contrast.

[0006]

According to the first aspect of the present invention,
An error between the output pixel data quantized for each pixel of the multi-valued pixel data input for each line and the input pixel data is diffused into peripheral pixel data, and the result is binarized based on a predetermined threshold value. In an image processing method of processing and outputting output binary pixel data, an all-black / all-white determination means for determining whether all pixels of the input multi-valued pixel data are all black or all white is provided, 3. The method according to claim 2, wherein an error from the input pixel data is set to "0" when a predetermined number of consecutive pieces of pixel data determined to be all black or all white, and when the pixel data is determined to be all black or all white. According to the invention, a result of diffusing an error between input pixel data and output pixel data quantized for each pixel of multi-valued pixel data input for each line into peripheral pixel data is based on a predetermined threshold value. Binary In an image processing method of processing and outputting output binary pixel data, an all-black / all-white determination means for determining whether all pixels of the input multi-valued pixel data are all black or all white is provided, When it is determined that the error is all black or all white and the diffused accumulated error is larger than a predetermined set value, the error with the input pixel data is set to “0”, and the error set all black or all white is set. The threshold value when the data is binarized is periodically changed for each pixel to be processed, and the set value for comparing the accumulated error is set to the maximum value of the threshold value for periodically changing the threshold value, and to all black or all white. Characterized by being less than the difference from the level determined as
According to a third aspect of the present invention, a result obtained by diffusing an error between output pixel data quantized for each pixel of input multi-valued pixel data for each line and input pixel data to peripheral pixel data is set to a predetermined value. In an image processing method for performing binarization processing based on a threshold value and outputting output binary pixel data, an all-black / all-white determination for determining whether each pixel of the input multi-valued pixel data is all black or all white Means for determining whether all the pixels are all black or all white, binarizing the pixel data as all black data or all white data regardless of an error with the input pixel data, 5. The method according to claim 4, wherein the threshold value at the time of performing the quantization processing is periodically changed for each processing pixel, and the output obtained by performing the quantization processing for each pixel of the multi-valued pixel data input for each line. Pixel Day In the image processing method of outputting an output binary pixel data by binarizing a result obtained by diffusing an error between the input pixel data and peripheral pixel data into peripheral pixel data based on a predetermined threshold value, All-black all-white discriminating means for discriminating all black or all white for each pixel of the value pixel data is provided, and when the all-black all-white discriminating means determines that neither all black nor all white, the threshold value is cycled. The threshold value is set to an all-white level for all-black data, and is set to an all-white level for all-black data. It is characterized by setting to all black levels,
The invention according to claim 5 is characterized in that the determination by the all-black / all-white determining means is performed only for either all black or all white.

[0007]

According to the first aspect of the present invention, a result obtained by diffusing an error between output pixel data quantized for each pixel of multi-valued pixel data input for each line and input pixel data into peripheral pixel data is defined as a predetermined value. 2 based on the set threshold
In an image processing method of performing binarization processing and outputting output binary pixel data, an all-black / all-white discriminating unit that determines whether each pixel of input multi-valued pixel data is all black or all white is provided. When each pixel is determined to be all black or all white, and when a predetermined number of all black or all white pixel data continue, an error from the input pixel data is set to “0”.

Accordingly, the contrast of a halftone image can be improved by setting the binarization error to "0" when all black or all white pixel data is input.

According to the second aspect of the present invention, a result obtained by diffusing an error between input pixel data and output pixel data quantized for each pixel of multi-valued pixel data input for each line into peripheral pixel data is determined by a predetermined value. In an image processing method of performing binarization processing based on a set threshold value and outputting output binary pixel data, it is determined whether all pixels of input multi-valued pixel data are all black or all white. White determining means is provided,
When it is determined that all black or all white for each pixel, and when the diffused accumulated error is larger than a predetermined set value,
An error from the input pixel data is set to “0”, and a threshold value for binarizing the all-black or all-white data set with the error is periodically changed for each processing pixel. The set value for comparing the error is set to be less than the difference between the maximum value of the threshold value that is periodically changed and the level determined as all black or all white.

Therefore, when all black or all white pixel data is input, the binarization error is set to "0" according to the accumulated error amount, so that the binarization threshold value is periodically set. Even if it is changed, the occurrence of white spots in a black background and black spots in a white background can be eliminated.

According to the third aspect of the present invention, a result obtained by diffusing an error between input pixel data and output pixel data quantized for each pixel of multi-valued pixel data input for each line into peripheral pixel data is defined as a predetermined value. In an image processing method of performing binarization processing based on a set threshold value and outputting output binary pixel data, it is determined whether all pixels of input multi-valued pixel data are all black or all white. White determining means is provided,
When it is determined that each pixel is all black or all white, the pixel data is binarized as all black data or all white data regardless of an error with the input pixel data, and the binarization processing is performed. The threshold value is periodically changed for each processing pixel.

Therefore, according to all black and all white data, 2
Since the data is converted into the coded data, the contrast of the halftone image can be improved while keeping the accumulated error.

According to the fourth aspect of the present invention, a result of diffusing an error between output pixel data quantized for each pixel of the multi-valued pixel data inputted for each line and input pixel data into peripheral pixel data is determined by a predetermined value. In an image processing method for performing binarization processing based on a set threshold value and outputting output binary pixel data, all black or all white for each pixel of the input multi-valued pixel data is determined. All-white determining means is provided, and when it is determined by the all-black and all-white determining means that the image is neither all black nor all white, the threshold value is periodically changed and the all black and all white determining means is used. When it is determined to be black or all white, the threshold value is set to an all white level for all black data and to an all black level for all white data.

Therefore, according to all black and all white data, 2
Since the binarized threshold value is changed, the contrast of the halftone image can be improved while maintaining the accumulated error.

In the fifth aspect of the present invention, the determination by the all-black / all-white determination means is performed only for either all black or all white.

[0016]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be specifically described below based on embodiments.

FIGS. 1 to 4 show an embodiment of a facsimile apparatus to which an image processing apparatus is applied.

FIG. 1 is a block diagram of a facsimile apparatus 1. The facsimile apparatus 1 includes an image sensor 2 and an image sensor A.
It comprises a / D converter 3, a timing controller 4, a RAM 5, a digital image processing unit 6, and the like.

The image sensor 2 converts the reflected light from the original into an electric signal and outputs a predetermined analog signal to the A / D converter 3. The A / D converter 3 is input from the image sensor 2. The analog signal is converted into a predetermined digital signal and output to the digital image processing unit 4.

The timing controller 4 outputs a control signal for controlling the operation timing of each unit to each unit,
(Random Access Memory) 5 holds and updates various kinds of reference data, and holds and updates error data necessary for error diffusion processing described later.

As shown in FIG. 2, the digital image processing section 6 comprises an error calculating section 7, a density correcting section 8, a binarizing section 9, a binarizing error generating section 10, and a latch section 11.
The error diffusion processing unit is configured as a whole.

The error calculator 7 calculates an error amount for the current pixel from the previous binarized error data held in the RAM 5 and outputs the calculation result to the density corrector 8.

The density correction unit 8 adds the error calculation result input from the error calculation unit 7 to the read data (digital signal) input from the A / D converter 3 and binarizes the density correction data. And outputs the result to the binarization error calculating unit 10.

The binarization unit 9 binarizes the density correction data input from the density correction unit 8 with a predetermined threshold value input thereto, and converts the binarized data to a binarization error generation unit 10 (not shown). Output.

The binarization error calculator 10 includes a density difference corrector 8
The binarization error in the pixel is calculated according to the binarization data input from the binarization unit 9 based on the density correction data input from the CPU 11 and output to the latch unit 11.

The latch unit 11 adjusts the output timing of the binarized error data input from the binarized error generator 10 and outputs the adjusted data to the RAM 5 and the error calculator 7.

Next, the operation will be described.

An analog signal corresponding to one line of the original image read by the image sensor 2 is converted into a digital signal by the A / D converter 3 and output to the digital image processing unit 6, where the analog signal is output. Error diffusion processing is performed for each pixel of one line. FIG. 3 shows an example of an error diffusion filter used in the error diffusion processing. FIG. 3 (a)
Indicates the distribution ratio of the current pixel * and its surrounding pixels.
(B) shows error data of the current pixel density.

[0029] Assuming that corrected density _{a'm, n} for the current pixel density a _{m, n,} the following calculation is performed. a ′ _{m, n} = am _{, n} + (2ε _{m, n−1} + 2ε _{m−1, n} + ε _{m−1, n−1}
+ Ε _{m-1, n + 1} ) / 6

That is, an error amount for the current pixel am _{, n} is calculated by the error calculator 7 based on the error data of the previous pixel of the previous line obtained by the binarization error generator 10 and held in the RAM 4. The error amount is calculated by the density correction unit 8 at the current pixel a.
_m, is added at a predetermined ratio to the _n density correction performed by the corrected density data _{a'm, after n} has been determined, the corrected density data _{a'm, n} are, binarization section 9 Binarization processing is performed.

Here, the binarized signal based on the corrected density data a ′ _{m, n} is represented by ad _{m, n} , and the threshold level of the binarization is represented by T
Assuming that H, when a ′ _{m, n} ≧ TH, ad _{m, n} = 1, ε _{m, n} = (a ′ _{m, n} −
M) When a ′ _{m, n} <TH, ad _{m, n} = 0, ε _{m, n} = a ′ _{m, n} where ε _{m, n} : a binarization error with respect to the current pixel density a _{m, n} M: Number of gradations (a ' _{m, n} -M): Complement of a' _{m, n} , and the number of gradations M is 16 and the total black level is 1
5, if the total white level is 0, a ' _{m, n} = 11, and TH = 8, ad _{m, n} = 1 (black), ε _{m, n} = 11-15 = -4
Becomes That is, by binarizing a pixel having a density of 11 into black (15), a density difference (error) of (-4) between input and output pixels is obtained.
Is generated, and this error is successively diffused to subsequent pixels for liquidation. Note that ε _{m, n} is calculated by the binarization error generation unit 10 and is stored in the RAM 5 at a predetermined timing via the latch unit 11.
Is forwarded to

However, if the threshold value at the time of binarization is kept constant, a pixel having a density near the threshold value cannot fully diffuse the error and a tailing (texture) phenomenon occurs.
This phenomenon can be eliminated by periodically changing the threshold value for each pixel.

In the case of an image processing apparatus that also shares a dither processing unit, the load on the hardware / software of the image processing apparatus can be reduced by sharing a threshold value setting register for dither. FIG. 4 shows a configuration example of the image processing apparatus 21 sharing the threshold value setting register.

In FIG. 4, an image processing device 21 comprises a multiplexer 22, a threshold value register 23, a system control unit 24, and a comparator 25. One of the signal and the error diffusion pixel signal input after error diffusion processing is selected and output to the comparator 25, and the threshold value is compared from the threshold value register (data table) 23 in which the threshold value for each pixel is set. When the dither pixel signal or the error diffusion pixel signal is output to the comparator 25 based on the threshold value input by the comparator 25, it is binarized and output as binary data. The system control unit 24 controls the operations of these units, and specifies a threshold value to be read from the threshold value register 23.

The pixel density of the read data is all black level (15), the accumulated error is (-6),
Assuming that the threshold value is changed in the range of 6 to 10,
Since the data after the density correction is 9, the threshold value is 6
-9, the binarized output is 1 (black), 2
The binarization generation error is stored as 15-9 = 6. When this state continues and the threshold value is set to 10, the binary output becomes 0 (white), and the output side is output as all white on the output side even though the input data is at all black levels. Further, even when the threshold value is not changed, white spots are likely to be generated around the black background portion, giving an impression that the contrast of the image is blurred.

FIGS. 6 to 9 show one embodiment of an image processing apparatus to which the image processing system according to the first to fifth aspects of the present invention is applied in order to eliminate such image blurring.

FIG. 6 is a block diagram of a main part of an image processing apparatus 31 having an error diffusion processing function. The same components as those in the image processing apparatus shown in FIG. Omitted.

In FIG. 6, an all-black / all-white judging section (all-black and all-white discriminating means) 32 judges whether the read data is all black or all white, for example, (1, 0) all black, (0) , 1) all white, (0,
0) The signals of non-all black and all white are converted to the binary error generators 9 and 2
Output to the value conversion unit 11.

The system control unit 33 controls the operation of each unit and inputs threshold value data for binarization to the binarization unit 11.

Next, the operation will be described.

The signals of (1,0) all black, (0,1) all white, (0,0) non-all black, and all white are output from the all black / all white determining unit 32 to the binarization error generating unit 9 and When output to the binarization unit 11, the binarization error is set to “0” in the case of all black or all white. In the binarization unit 11, when all black or all white is detected, system control is performed. Regardless of the threshold value input from the unit 33, the binarized data is output as black or white.
This binarized output may be obtained by converting the data after binarization, or by setting the threshold level to an all white level for all black data and to an all black level for all white data. Is also good.

The correction for all black and all white may not be performed when the accumulated error is small. That is, when the cumulative error is equal to or more than (−5), 6 to 1
It does not affect a threshold value of zero. Further, the determination of all black and all white may be made by performing only one of the determinations.

FIGS. 6 to 8 show specific circuit configuration examples for processing the determination of all black and all white.

FIG. 6 shows a logic circuit configuration applied to the binarization error generating section 9 of FIG. 5, and includes a plurality of pixels input from the density correcting section 8 in response to all black and all white inputs. The binarization errors of the density data of
5 and output to the error calculator 6.

FIG. 7 shows a logic circuit configuration applied to the binarization unit 11 of FIG. 5, and not the threshold value input from the system control unit 33 but the all black / all white determination unit 32.
The density data input from the density correction unit 8 is compared with the all black and all white data input from the CPU, and the binarized data is output.

FIG. 8 shows a logic circuit configuration of a portion for changing the threshold value applied to the binarizing section 11 of FIG. The threshold value input from the system control unit 33 is selected according to the white data, the threshold value is output, and the binarization operation is performed in the binarization unit 11 (not shown).

FIG. 9 shows a flowchart of the binarization processing procedure involved in the above all black and all white determination processing in the image processing apparatus 31.

In FIG. 9, it is determined whether the data input from the all black / all white determining section 32 is all black or all white (step S).
1) In the case of all black or all white, in the processing according to the first aspect of the present invention, the binarization error is set to “0” (step S).
2) In the processing according to the second aspect of the present invention, the threshold value is converted according to the accumulated error (step S3), the binarization processing is performed (step S4), and the binarization error is calculated (step S5). In the processing according to the third or fourth aspect of the present invention, the data is changed to black or white binary data regardless of the binarization error (step S6), and the binarization error is calculated (step S7).

On the other hand, if it is not all black or all white in step Sl, a density correction operation is performed to obtain corrected density data (step S8), and the corrected density data is input to the threshold value input from the system controller 33. And binarize to 2
In addition to outputting the value data, a binarization error is calculated using the corrected density data (step S10).

Therefore, the contrast of the halftone image can be improved by setting the binarization error in the case of inputting all black or all white pixel data to "0". When the data is input, the binarization error is set to “0” according to the accumulated error amount. Therefore, even if the threshold value of the binarization is periodically changed, the white point in the black background or the white background , And conversion to binary data according to all black and all white data, it is possible to improve the contrast of a halftone image while preserving the accumulated error.

When calculating the binarization error, there is a case where a fraction is rounded. If the fraction is rounded in a direction approaching “0”, the integer conversion of the error can be expedited. Can be improved. In addition, when the result of the two's complement operation when calculating the binarization error is a negative value, the error is converted to an integer by adding “1” after rounding the fraction. And the image quality can be improved earlier.

For example, when −11 ÷ 2 = −5.5, 00101 (two's complement) is shifted by 1 bit to 0
010 (−6) is increased by 1 and corrected to (−5).

Therefore, since the fraction of the error calculation is converted to an integer in a direction approaching "0", the conversion of the error into an integer is quick and a sharp halftone image can be reproduced.

[0054]

According to the first aspect of the present invention, an error between output pixel data quantized for each pixel of multi-valued pixel data input for each line and input pixel data is diffused to peripheral pixel data. In an image processing method in which a result is binarized based on a predetermined threshold value and output binary pixel data is output, it is determined whether each pixel of input multi-valued pixel data is all black or all white. An all-black-all-white discriminating means is provided, and when the pixel is determined to be all-black or all-white, when an all-black or all-white pixel data continues for a predetermined number, the error from the input pixel data is set to “0”. Since the setting is performed, the contrast of the halftone image can be improved by setting the binarization error to “0” when all black or all white pixel data is input.

According to the second aspect of the present invention, a result obtained by diffusing an error between input pixel data and output pixel data quantized for each pixel of multi-valued pixel data input for each line into peripheral pixel data is obtained. In an image processing method in which binarization processing is performed based on a predetermined set threshold value and output binary pixel data is output, all black or white is determined for each pixel of input multi-valued pixel data. A black-and-white determining means is provided,
When it is determined that each pixel is all black or all white, and when the accumulated error that has been diffused becomes larger than a predetermined set value, the error from the input pixel data is set to “0”, and the error-set all black is set. Alternatively, the threshold value for binarizing the all-white data is periodically changed for each processing pixel, and the set value for comparing the accumulated error is set to the maximum value of the periodically changed threshold value and the all-black or Since the value is set to be less than the difference from the level determined to be all white, when the pixel data of all black or all white is input, the binarization error is set to “0” according to the accumulated error amount. Therefore, even if the threshold value for binarization is changed periodically, it is possible to eliminate the occurrence of white points in a black background and black points in a white background.

According to the third aspect of the present invention, the result of diffusing the error between the input pixel data and the output pixel data quantized for each pixel of the multi-valued pixel data input for each line into the peripheral pixel data is obtained. In an image processing method in which binarization processing is performed based on a predetermined set threshold value and output binary pixel data is output, all black or white is determined for each pixel of input multi-valued pixel data. A black-and-white determining means is provided,
When it is determined that each pixel is all black or all white, the pixel data is binarized as all black data or all white data regardless of an error with the input pixel data,
Since the threshold value at the time of the binarization processing is periodically changed for each processing pixel, the threshold value is converted into binary data according to all black and all white data. Image contrast can be improved.

According to the fourth aspect of the present invention, a result obtained by diffusing an error between output pixel data quantized for each pixel of multi-valued pixel data inputted for each line and input pixel data into peripheral pixel data is obtained. In an image processing method of performing binarization processing based on a predetermined set threshold value and outputting output binary pixel data, it is determined whether each pixel of the input multi-valued pixel data is all black or all white. When all-black / all-white discriminating means is provided and the all-black / all-white discriminating means determines that the image is neither all-black nor all-white, the threshold value is periodically changed, and the all-black / all-white discriminating means is used. Alternatively, when the threshold value is determined to be all white, the threshold value is set to an all white level for all black data and to an all black level for all white data. Binarization thread Since changing the Interview value can improve the contrast of the halftone image while preserving the accumulated error.

According to the fifth aspect of the present invention, the determination by the all-black and all-white determination means may be performed only for either all black or all white.

[Brief description of the drawings]

FIG. 1 is a block diagram of a facsimile apparatus to which an image processing apparatus is applied.

FIG. 2 is a block diagram of a digital image processing unit in FIG. 1;

FIG. 3 is a diagram showing an example of an error diffusion filter used at the time of error diffusion processing in the digital image processing unit in FIG. 2;

FIG. 4 is a block diagram of an image processing apparatus.

FIG. 5 is a block diagram of a main part of the image processing apparatus according to the first to fifth aspects of the invention

FIG. 6 is a diagram showing a configuration of a logic circuit applied to a binarization error generation unit in FIG. 5;

FIG. 7 is a diagram showing a configuration of a logic circuit applied to the binarization unit in FIG. 5;

FIG. 8 is a diagram showing a logic circuit configuration of a part that changes a threshold value applied to the binarization unit in FIG. 5;

FIG. 9 is a flowchart of an error diffusion process according to the present invention.

[Explanation of symbols]

 Reference Signs List 1 facsimile device 5 RAM 6 digital image processing unit 7 error calculation unit 8 density correction unit 9 binarization unit 10 binarization error generation unit 21 image processing unit 22 multiplexer 23 threshold value register 24 system control unit 25 comparator 31 image processing Device 32 All black / all white judgment unit 33 System control unit

Claims (5)

[Claims] 1. A method according to claim 1, wherein an error between the output pixel data quantized for each pixel of the multi-valued pixel data input for each line and the input pixel data is diffused into peripheral pixel data to obtain a predetermined threshold value. In an image processing method of performing binarization processing based on the output and outputting output binary pixel data, an all-black / all-white determining means for determining whether all pixels of the input multi-valued pixel data are all black or all white is provided. When it is determined that all black or all white for each pixel, when a predetermined number of consecutive pixel data determined to be all black or all white,
An image processing method, wherein an error from the input pixel data is set to "0". 2. The method according to claim 1, wherein an error between the output pixel data quantized for each pixel of the multi-valued pixel data input for each line and the input pixel data is diffused to peripheral pixel data to obtain a predetermined threshold value. In an image processing method of performing binarization processing based on the output and outputting output binary pixel data, an all-black / all-white determining means for determining whether all pixels of the input multi-valued pixel data are all black or all white is provided. When each pixel is determined to be all black or all white, and when the diffused accumulated error becomes larger than a predetermined set value, an error from the input pixel data is set to “0” and the error is set. The threshold value for binarizing all black or all white data is periodically changed for each processing pixel, and a set value for comparing the accumulated error is set to the maximum value of the threshold value for periodically changing the threshold value. All black or all white The image processing method characterized by less than the difference between the determined the level. 3. A predetermined threshold value obtained by diffusing an error between output pixel data quantized for each pixel of input multi-valued pixel data and input pixel data into peripheral pixel data into a predetermined threshold value. In an image processing method of performing binarization processing based on the output and outputting output binary pixel data, an all-black / all-white determining means for determining whether all pixels of the input multi-valued pixel data are all black or all white is provided. When it is determined that each pixel is all black or all white, the pixel data is binarized as all black data or all white data and binarized regardless of an error with the input pixel data. Wherein the threshold value is periodically changed for each processing pixel. 4. A predetermined threshold value is obtained by diffusing an error between output pixel data and input pixel data, which are quantized for each pixel of multi-valued pixel data input for each line, into peripheral pixel data. In an image processing method of performing binarization processing based on the output and outputting output binary pixel data, an all-black / all-white determining means for determining whether all pixels of the input multi-valued pixel data are all black or all white is provided. When it is determined by the all-black-all-white discriminating means that the image is neither all black nor all-white, the threshold value is periodically changed, and when the all-black all-white discriminating means determines that the image is all black or all white, An image processing method, wherein the threshold value is set to an all white level for all black data and to an all black level for all white data. 5. The image processing method according to claim 1, wherein the determination by said all-black / all-white determining means is performed only for one of all black and all white.