JP3549475B2 - Image processing method - Google Patents

Description

[0001]
[Industrial applications]
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]
[Prior art]
2. Description of the Related Art Recently, in an information 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 quantized for each pixel of multi-valued pixel data and input 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]
[Problems to be solved by the invention]
However, in such a conventional error diffusion method, when the input pixel data is at an all black level or an all white level, the accumulated error of the pixel data up to that point does not occur even though the binarization error does not occur. If the threshold value of the binarization is always kept constant and stored, a texture peculiar to error diffusion (an image trailing phenomenon that occurs when error diffusion is not sufficient) occurs. However, there is a problem that 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.
[0004]
In a conventional error diffusion process performed by an image processing apparatus described in, for example, Japanese Patent Application Laid-Open No. 1-284172, the remainder (following the decimal point) of an error generated when a binarization error is obtained by weighting calculation ) Is added to the next pixel data to be correlated, corrected, the corrected pixel data is binarized by a certain threshold value, and the input image density and the output image density are stored to reproduce the image. However, this apparatus cannot remove the above-described texture unique to error diffusion.
[0005]
The present invention is, when performing the error diffusion processing, when it is determined whether the all-black or all white for each pixel, a threshold value for the all-black data, the all white level with respect to the total white data Is to reduce the occurrence of white points in a black background and the occurrence of black points in a white background by binarizing pixel data set to all black levels with a threshold value periodically changed for each processing pixel, It is an object of the present invention to provide an image processing method for improving the contrast of a halftone image.
[0006]
[Means for Solving the Problems]
According to the first 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 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 all pixels of the input multi-valued pixel data are all black or all white means provided, when it is determined that neither all white at all black by previous SL all black all white determining means, said thresh values with periodically changing, and full black or all-white by the all black all white determining means When determined, the threshold value is set to an all-white level for all black data, and set to an all-black level for all white data,
The invention according to claim 2 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]
[Action]
[0008]
[0009]
[0010]
[0011]
[0012]
[0013]
According to the first aspect of the present invention, a result of 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 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 all pixels of the input multi-valued pixel data are all black or all white Means is provided, and 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 the all black or all-white discriminating means is used. When determined to be 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.
[0014]
Therefore, since the binarized threshold value is changed according to the all-black and all-white data, the contrast of the halftone image can be improved while keeping the accumulated error.
[0015]
According to the second aspect of the present invention, the determination by the all-black / all-white determining means is performed only for all black or all white.
[0016]
【Example】
Hereinafter, the present invention will be specifically described based on examples.
[0017]
FIGS. 1 to 4 are diagrams showing an embodiment of a facsimile apparatus to which an image processing apparatus is applied.
[0018]
FIG. 1 is a block diagram of the facsimile apparatus 1. The facsimile apparatus 1 includes an image sensor 2, an A / D converter 3, a timing controller 4, a RAM 5, a digital image processing unit 6, and the like.
[0019]
The image sensor 2 converts reflected light from the document into an electric signal and outputs a predetermined analog signal to the A / D converter 3. The A / D converter 3 converts the analog signal input from the image sensor 2 into an electric signal. The signal is converted into a predetermined digital signal and output to the digital image processing unit 4.
[0020]
The timing controller 4 outputs a control signal for controlling the operation timing of each section to each section, and a RAM (Random Access Memory) 5 holds and updates various reference data, and stores error data necessary for error diffusion processing described later. Retain and update.
[0021]
As shown in FIG. 2, the digital image processing unit 6 includes an error calculation unit 7, a density correction unit 8, a binarization unit 9, a binarization error generation unit 10, and a latch unit 11. A diffusion processing unit is configured.
[0022]
The error calculator 7 calculates an error amount for the current pixel based on the previous binarized error data held in the RAM 5 and outputs the calculation result to the density corrector 8.
[0023]
The density correction unit 8 adds the error amount calculation result input from the error calculation unit 7 to the read data (digital signal) input from the A / D converter 3, and converts the density correction data into a binarization unit 9. The output is output to the binarization error calculator 10.
[0024]
The binarization unit 9 binarizes the density correction data input from the density correction unit 8 with a predetermined threshold value input, and outputs the binarized data to the binarization error generation unit 10 outside the figure.
[0025]
The binarization error calculation unit 10 calculates a binarization error in the pixel according to the binarization data input from the binarization unit 9 based on the density correction data input from the density difference correction unit 8 and latches it. Output to the unit 11.
[0026]
The latch unit 11 adjusts the output timing of the binarized error data input from the binarized error generation unit 10 and outputs the adjusted data to the RAM 5 and the error calculation unit 7.
[0027]
Next, the operation will be described.
[0028]
An analog signal for 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 one line of the analog image is output. An error diffusion process is performed for each pixel. FIG. 3 shows an example of an error diffusion filter used in the error diffusion processing. FIG. 3A shows the distribution ratio of the current pixel * and its surrounding pixels, and FIG. 3B shows error data of the current pixel density.
[0029]
Now, the current pixel density a _m, after correction for _n concentration _a'm, when the _n, the following calculation is performed.
a ′ _{m, n} = am _{, n} + (2εm _{, n−1} + 2εm _{−1, n} + εm _{−1, n−1} + εm _{−1, n + 1} ) / 6
[0030]
That is, an error amount for the current pixel am _{, n} is calculated by the error calculation unit 7 based on the error data of the previous pixel of the previous line obtained in the binarization error generation unit 10 and held in the RAM 4, and the error amount is calculated. Is added to the current pixels am _{, n} at a predetermined ratio by the density correction unit 8 to perform density correction to obtain corrected density data a ′ _{m, n. Then} , the corrected density data a ′ _{m, n} is binarized by the binarization unit 9.
[0031]
Here _{, assuming} that the binarized signal based on the corrected density data a ′ _{m, n} is ad _{m, n} and the threshold level of the binarization is TH,
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}
Here, ε _{m, n} : binarization error M with respect to current pixel density am _{, n} : gradation number (a ′ _{m, n} −M): complement of a ′ _{m, n} , and gradation number M is now If the total black level is 15, the total white level is 0, and a ′ _{m, n} = 11 and TH = 8 in 16 steps, ad _{m, n} = 1 (black), ε _{m, n} = 11−15 = -4. That is, the binarization of the pixel having the density of 11 into black (15) causes a density difference (error) of (-4) between the input and output pixels, and this error is sequentially diffused to the subsequent pixels for liquidation. I will do it. Note that ε _{m, n} is calculated by the binarization error generation unit 10 and transferred to the RAM 5 at a predetermined timing via the latch unit 11.
[0032]
However, if the threshold value at the time of binarization is made constant, a pixel having a density near the threshold value cannot fully diffuse an error and a tailing (texture) phenomenon occurs. It can be removed by changing it periodically.
[0033]
Further, in the case of an image processing apparatus that also shares a dither processing unit, the load on 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.
[0034]
4, the image processing device 21 includes a multiplexer 22, a threshold value register 23, a system control unit 24, and a comparator 25. The dither pixel signal input after being dither-processed by the multiplexer 22 out of the figure is input. One of the error-diffusion pixel signals input after being subjected to the diffusion process is selected and output to the comparator 25, and the threshold value is sent to the comparator 25 from a threshold value register (data table) 23 in which a threshold value for each pixel is set. When output, the dither pixel signal or the error diffusion pixel signal is binarized based on the threshold value input by the comparator 25 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.
[0035]
Now, assuming that the pixel density of the read data is all black level (15), the accumulated error is (-6), and the threshold value is changed in the range of 6 to 10, the data after density correction is When the threshold value changes from 6 to 9, the binarized output is stored as 1 (black) and the binarization generation error is 15-9 = 6. When this state continues and the threshold value is set to 10, the binarized output becomes 0 (white), and the output side is output as all white even though the input data is all black. Further, even when the threshold value is not changed, a white point is likely to be generated around a black background portion, giving an impression that the contrast of an image is blurred.
[0036]
FIGS. 6 to 9 show an embodiment of an image processing apparatus to which the image processing method according to the first and second aspects of the present invention is applied in order to eliminate such image blurring.
[0037]
FIG. 6 is a block diagram of a main part of the image processing apparatus 31 having the error diffusion processing function. The same components as those of the image processing apparatus shown in FIG.
[0038]
In FIG. 6, an all-black / all-white determining unit (all-black and all-white determining unit) 32 determines whether the read data is all black or all white. For example, (1, 0) all black, (0, 1) A signal of all white, (0,0) non-all black, and all white is output to the binarization error generator 9 and the binarizer 11.
[0039]
The system control unit 33 controls the operation of each unit and inputs threshold value data for binarization to the binarization unit 11.
[0040]
Next, the operation will be described.
[0041]
A signal of (1,0) all black, (0,1) all white, (0,0) non-all black, and all white is sent from the all black / all white determination unit 32 to the binarization error generation unit 9 and the binarization signal. When output to the unit 11, the binarization error is set to “0” in the case of all black or all white. When the binarization unit 11 detects all black or all white, the system control unit 33 outputs The binarized data is output as black or white regardless of the input threshold value. 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.
[0042]
In addition, 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), the threshold value of 6 to 10 is not applied. Further, the determination of all black and all white may be performed by only one of the determinations.
[0043]
6 to 8 show specific circuit configuration examples for processing the determination of all black and all white.
[0044]
FIG. 6 shows a logic circuit configuration applied to the binarization error generation unit 9 of FIG. 5, and the density data of a plurality of pixels input from the density correction unit 8 in response to the input of all black and all white. Are calculated and output to the RAM 5 and the error calculator 6 (not shown).
[0045]
FIG. 7 shows a logic circuit configuration applied to the binarization unit 11 of FIG. 5. The logic circuit configuration is not a threshold value input from the system control unit 33 but a total value input from the all black / all white determination unit 32. The density data input from the density correction unit 8 is compared with the black and all white data, and is binarized to output binarized data.
[0046]
FIG. 8 shows a logical circuit configuration of a portion that changes the threshold value applied to the binarization unit 11 in FIG. 5, and converts all-black and all-white data input from the all-black / all-white determination unit 32. In response, the threshold value input from system control unit 33 is selected, the threshold value is output, and a binarization operation is performed in binarization unit 11 (not shown).
[0047]
FIG. 9 shows a flowchart of a binarization processing procedure involved in the above all black and all white determination processing in the image processing apparatus 31.
[0048]
9, data input from the full black / all-white judgment unit 32 determines whether all black or all white (step Sl), when the all-black or all-white, in the processing of one example, the binarization error Is set to "0" (step S2). In another example of the process, the threshold value is converted according to the accumulated error (step S3), the binarization process is performed (step S4), and the binarization error is calculated (step S4). In step S5), in the processing according to the first aspect of the invention, the binary data is changed to black or white binary data regardless of the binary error (step S6), and the binary error is calculated (step S7).
[0049]
If it is not all black or all white in step Sl, density correction calculation is performed to obtain corrected density data (step S8), and the corrected density data is converted into a binary value by a threshold value input from the system control unit 33. The binarization process is performed to output binary data, and a binarization error is calculated using the corrected density data (step S10).
[0050]
Therefore, the contrast of the halftone image can be improved by setting the binarization error to “0” when the all-black or all-white pixel data is input, and the all-black or all-white pixel data is input. In this case, 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 black point in the white background is not changed. Since the occurrence can be eliminated and the data is converted into the binary data according to the all black and all white data, the contrast of the halftone image can be improved while keeping the accumulated error.
[0051]
In addition, when calculating a binarization error, a fraction may be rounded in some cases. If the fraction is rounded in a direction approaching “0”, the conversion of the error into an integer can be expedited, and the image quality can be improved. Can be planned. In addition, when the result of the two's complement operation when calculating the binarization error becomes a negative value, the error is converted to an integer by adding "1" after rounding the fraction. , And the image quality can be improved.
[0052]
For example, when −11 ÷ 2 = −5.5, 00101 (two's complement) is shifted by 1 bit, and 0010 (−6) is increased by 1 to be corrected to (−5).
[0053]
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]
【The invention's effect】
[0055]
[0056]
[0057]
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 to peripheral pixel data is set to a predetermined value. In an image processing method for performing binarization processing based on the threshold value thus obtained and outputting output binary pixel data, it is possible to determine whether each pixel of the input multi-valued pixel data is all black or all white. A white discriminating means is provided, and when it is judged by the all-black and all-white discriminating means that neither the black nor the all-white, the threshold value is periodically changed and the all-black or all-white discriminating means is used. When the threshold value is determined, the threshold value is set to an all-white level for all-black data and to an all-black level for all-white data. Change threshold value Since thereby it is possible to improve the contrast of the halftone image while preserving the accumulated error.
[0058]
According to the second aspect of the present invention, the determination by the all-black / all-white determination means may be performed for only one of all black and all white.
[Brief description of the drawings]
FIG. 1 is a block diagram of a facsimile apparatus to which an image processing device is applied. FIG. 2 is a block diagram of a digital image processing unit in FIG. 1. FIG. 3 is used in the digital image processing unit in FIG. block diagram of Figure 4 shows an image processing apparatus showing an example of an error diffusion filter [5] binarization error of claim 1, 2 main block diagram of an image processing apparatus according to the invention described [6] 5 FIG. 7 is a diagram showing a logic circuit configuration applied to a generation unit; FIG. 7 is a diagram showing a logic circuit configuration applied to a binarization unit in FIG. 5; FIG. 8 is a threshold value applied to the binarization unit in FIG. FIG. 9 is a diagram showing a logic circuit configuration of a part for changing the error. FIG. 9 is a flowchart of an error diffusion process according to the inventions described in claims 1 and 2.
1 Facsimile machine 5 RAM
Reference Signs List 6 digital image processing unit 7 error calculation unit 8 density correction unit 9 binarization unit 10 binarization error generation unit 21 image processing device 22 multiplexer 23 threshold value register 24 system control unit 25 comparator 31 image processing device 32 All white determination unit 33 System control unit

Claims (2)

A result obtained by diffusing an error between output pixel data, which is quantized for each pixel of the multi-valued pixel data input for each line, and input pixel data into peripheral pixel data is binarized based on a predetermined threshold value. In an image processing method of processing and outputting output binary pixel data,
All-black all-white discriminating means for discriminating all black or all white for each pixel of the input multi-valued pixel data is provided,
When it is determined by the all-black-all-white discriminating means that neither all black nor all-white, the threshold value is periodically changed, and when it is discriminated as all black or all-white by the all-black all-white discriminating means, An image processing method characterized in that a threshold value is set to an all white level for all black data and to an all black level for all white data . 2. The image processing method according to claim 1, wherein the determination by the all-black / all-white determination means is performed only for one of all black and all white.

Images