JPS6356063A - Picture processing method - Google Patents

Abstract

PURPOSE:To identify a fogged character included in a character area or a character of a low density as a character properly by distinguishing a character picture aud an intermediate tone picture based on the ratio of a contrast level of a noted picture element to a mean value of a contrast level of surrounding picture elements. CONSTITUTION:A filter processing circuit 4 receives a density data of picture elements around a noted picture element from a picture memory 2 to obtain an adjacent picture element mean density DaV. An arithmetic circuit 5 obtains a ratio Dcn/Dav of the mean density Dav to the density data Dcn of the noted picture element inputted from the picture element memory 2, a deciding circuit 6 compares the ratio Dcn/Dav with a threshold value and outputs a decision signal SL. An arithmetic circuit 7 uses the adjacent picture element mean density Dav and a center picture element density Dcn to attain the arithmetic operation and to output an arithmetic density Do. A multiplexer 8 selects the density Do according to the decision signal SL, a screen generating section 9 compares the storage data in the dither matrix with the input density data from the multiplexer 8 and outputs the result of comparison to a picture output device 10.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention provides a method for reading a document containing a mixture of binary images such as characters and halftone images such as photographs, and when recording and reproducing the document.
The present invention relates to an image processing method that properly identifies seed images, applies appropriate processing to each image region, and obtains high-quality output.

[Prior Art] Many general document manuscripts contain a mixture of text and photographs, and when such manuscripts are recorded in binary format, the text is good, but the image quality of the photographs deteriorates, and On the other hand, if the image is recorded using a halftone generation method such as dither processing, the photograph will be good, but the image quality of the text will deteriorate. Therefore, in order to improve the image quality of both, for example, to simply binarize a character image and dither process a photographic image, it is necessary to properly distinguish between these two types of image areas. .

For this reason, in the prior art disclosed in Japanese Patent Laid-Open No. 58-220563 and Japanese Patent Laid-Open No. 58-220564, the sharp signal Dsh, which is the density data of the pixel of interest, is
and the unsharp signal [) ush, which is the average density value of the pixel of interest and pixels surrounding the pixel of interest, l
Take Dush −Dshl and set this to some threshold °[
When it exceeds h, it is determined to be a character area.

As shown in Figure 5, there is little change in the fA degree level when looking at the halftone area with sharp and unsharp signals (see Figure 5 (aHb)), but when looking at the text area with sharp and unsharp signals. This method focuses on the difference in concentration gradient, which shows that the change in concentration level is large when observed using a sharp signal (see FIG. 5 (cHd)).

[Problems to be Solved by the Invention] However, in this conventional method, area determination is performed based on whether the absolute value l Du, sh −Dshl of the difference between the sharp signal and the unsharp signal is greater than a certain threshold Th. This method is effective when the density of the text is very high and clear, but in reality, such ideal characters are rare, and there are usually many blurry characters with low density. , such conventional methods result in considerable judgment errors. Furthermore, even if the character image is close to the ideal state to some extent, due to the characteristics of the reading system, the character image loses a considerable amount of sharpness after being read, and this is also a cause of judgment errors in conventional devices. Become. FIG. 6 shows such a state, and is shown in FIG. 6(a).

(b) and (C) show the sharp signal Dsh, the unsharp signal Dush, and the absolute value of the difference lDush-Dshl for ideal clear characters, and (d) and (e) and (f) are Dsh and Dush for blurred characters obtained through the reading device, respectively.
and l[)ush-Dshl are shown. 6th
As is clear from comparing Figures (c) and (f), the absolute value I Dush - Dstl of the difference between the sharp signal Dsh and the unsharp signal Dush increases in the case of sharp characters, but The value is small in the case of a blurred character, and for this reason, in conventional devices, blurred character portions were often erroneously determined to be halftone areas. It is thought that this problem can be solved by changing the threshold 1aTh according to the degree of deterioration of the character image, but in reality, such processing is almost impossible, and even if it were realized, it would be a very complicated process. is likely to be necessary.

The present invention has been made in view of the above-mentioned circumstances, and it is an object of the present invention to provide an image processing method that can properly identify character data with low tJ degree or blur as characters. Furthermore, the present invention provides an image processing method that can obtain high-quality output images for both character areas and intermediate areas by accurately identifying areas, removing noise in halftone image areas, and emphasizing edges. This is what we are trying to provide.

[Means and effects for solving the problem] In this invention, a pixel matrix is set with the pixel of interest as the center,
The ratio between the gray level of the pixel of interest and the average gray level of the pixels in the pixel matrix is calculated, the calculated ratio is compared with a predetermined threshold, and based on the comparison result, the pixel of interest is determined to be a character image or a halftone image. to determine which of the following it is included in.

By taking the above ratio, the difference in level between clear characters and blurred characters becomes small. Therefore, by setting an appropriate predetermined rJJ value, not only clear characters but also blurred characters can be accurately identified without misjudgment. be able to.

Further, in the present invention, a pixel matrix that covers the pixel of interest is set as the center, the ratio of the gray level of the pixel of interest to the average gray level of the pixels in the pixel matrix is determined, and the calculated ratio is divided into two different It is compared with the first and second threshold values, and if the ratio ≧ the first threshold value, it is determined that it is a character area, and the first
If the ratio is equal to or greater than the second threshold, it is determined that the area is the outline of the halftone area, and if the ratio is the second threshold, it is determined that the area is other than the outline of the halftone area. If it is the first threshold, set the output gray level of the pixel of interest to either the maximum value or the minimum value that the gray level can take, and if the ratio of the first threshold > the second threshold, then the output gray level of the pixel of interest The output gradation level is set to a value obtained by contour enhancement processing according to a predetermined arithmetic expression, and when the output gradation level is the second threshold value, the output gradation level is set to the average gradation level of the pixels in the pixel matrix.

[Embodiment] FIG. 1 shows an example of an apparatus configuration when implementing the image processing method according to the present invention.

In FIG. 1, an image input device 1 includes, for example, a CC/D line image sensor, an A/D converter, etc.
The original image data read by the OD image sensor is converted into, for example, 8-bit multi-value density data (or reflectance data) by an A/D converter and output. The image memory 2 is, for example, a page memory, and stores multilevel density data input from the image input device @1 for one page of the original.

The image processing device 3 includes a filter processing circuit 4, an arithmetic circuit 5,
It is composed of a determination circuit 6, an arithmetic circuit 7, a multiplexer 8, and a screen generation section 9, and these components perform area determination processing, edge correction, dither processing, etc.
A density signal consisting of a 2Vi signal of "1" or "0" is output. The image input device 10 is, for example, a laser beam printer, a thermal printer, or an inkjet printer. If it is a laser beam printer, image recording is performed by turning the laser beam on and off according to a binary density signal input from the image processing device 3. Do the following.

Next, before explaining the configuration of the image processing device @3, the image processing device @3 will be described.
The processing procedure by the processing device will be explained in detail.

First, as shown in FIG. 2, consider a 3×3 pixel matrix centered on the pixel of interest Pcn. In this embodiment, in order to improve the degree of discrimination W of blurred characters, the ratio D sh of the sharp signal Dsh and the unsharp signal [) ush is
/Dush, and the character area and halftone area are identified based on this ratio [)sh/Qush.

This ratio of sharp and unsharp signals D sh/D
Since ush falls within a narrow range of values regardless of the condition of the original image, it is easier to set thresholds for area identification than before, and even for characters with severe image quality deterioration or characters with low temperature, it is only necessary to set a predetermined threshold. can be accurately identified. Incidentally, if the density of the pixel of interest Pcn is [)cn, and the 'a degrees of the surrounding pixels P1 to P8 are D1 to D8, respectively,
The sharp signal Dsh and the unsharp signal Dush were determined according to the following equations.

D sh = D cn = [) aV In other words, in this embodiment, in order to increase the degree of discrimination of character areas, the unsharp signal () ush is the average f of each pixel P1 to P8 in the pixel matrix excluding the pixel of interest Pcn:
It is calculated as the ja value.

FIG. 3 shows an example of the filter coefficients. In the conventional device, the unsharp signal [) ush is the image of interest 1
The average density value is obtained by appropriately weighting each pixel in the pixel matrix including Pcn.

FIG. 4 shows the ratio D of the sharp signal DCn and the unsharp signal [)aV for each of clear characters and blurred characters.
This shows an example of cn/D av.
The level difference between clear characters and blurred characters is smaller than before, so by setting an appropriate threshold Th,
Not only ideal characters but also blurred characters can be identified without misjudgment.

Now, if the ratio between the sharp signal and the unsharp signal is r, this ratio r will be r = 1 in areas where the density changes are gradual, such as halftones, and r = 1 in areas where the density changes are steep, such as text areas. 1.3 or more, and in the contour part of the halftone area, r
= takes a value of about 1.1 to 1.3.

Therefore, r for identifying character areas and halftone areas is
The JJ value - rh may be set to a value of about 1.3 to 1.5.

By the way, in this embodiment, the emphasis correction of the contour part in the halftone area and the smoothing process in the halftone area are performed based on the ratio r of the sharp signal and the unsharp signal obtained above using the above conditional expression (i) ( iiHiii).

(i) When ≧Th1 [)0 - maximum value or minimum value (ii) When Tt+1>r>Th2, Do = a (Dcn -Dav) + Dcn(i
ii) Do = Dav where r < Th2 In the above conditional expression, the first threshold value h1 is 1.3 to
1.5, the second threshold Th2 is a value close to 1, DO is the output pixel density, a is the edge enhancement coefficient, and the "maximum value" and "minimum value" are the input values, respectively. These are the highest and lowest concentration levels that the concentration data can take.

That is, in the case of (i) in the above conditional expression, it is determined that the pixel of interest Pcn is included in the character area, and its output density DO is set to the highest density level or the lowest density level.

In the case of (11), it is determined that the pixel of interest Pcn is included in the edge part of the halftone area, and the formula a (Dcn −Dav
)+Dcn, the edge portion is emphasized by determining the output density Qo.

In the case of (iii), it is determined that the pixel of interest pan is included in a part other than the edge of the halftone area, and the input density [)cn
By replacing and outputting the average value [)aV of surrounding pixels, smoothing is performed and noise that occurs when a high-resolution halftone image is input is removed.

Note that this average value [)ay may be either one excluding the pixel of interest or one including the pixel of interest.

Next, an example of the internal configuration of the image processing device 3 that performs the above operation will be shown.

When the density data Dcn of the pixel of interest Pcn is input from the image memory 2 to the arithmetic circuit 7, the filter 98 logic circuit 4 receives the surface tree P around the pixel of interest Pcn from the image memory 2.
Density data D1 to D8 of numbers 1 to P8 are input. In the filter processing circuit 4, the input density data D1
By applying the filter shown in FIG. 3 to D8 to D8, the average density [)aV of the adjacent pixels is inputted to the calculation circuit 5 and the calculation circuit 7. The arithmetic circuit 5 calculates the input average concentration [)
The ratio D cn/D av between aV and the density data [ ) cn of the pixel of interest inputted from the image memory 2 is determined, and the value D cn/D aV is inputted to the determination circuit 6 . The determination circuit 6 compares the input Uni D cn/D ay with the aforementioned two threshold values Th1 and Th2, and determines which of the aforementioned conditional expressions (i) and (iii) the pixel of interest Pcn belongs to. is determined, and the determined signal SL is input to the multiplexer 8.

In parallel with this processing, the Pf4g circuit 7 receives input adjacent pixel average density value [)ay and center pixel density [)cn.
, the calculations of the above-mentioned conditional expressions (iHii) and (iii) are performed in parallel, and each calculation value is input to the multiplexer 8.

That is, if the concentration level of the input data [)cn is, for example, O
If there are 256 levels from ~255, (1)
In the calculation, when the input data Dcn is determined to be a black pixel by comparing it with a predetermined threshold value, the input data Dcn
is replaced with the highest moisture level r255J, and when the pixel is determined to be white, the input data Dcn is replaced with the lowest moisture level r255J.
Replace it with OJ and output it. This is to prevent erroneously inverted output of black and white pixels in the subsequent dither processing in the screen generation section 9.

In addition, in the calculation of (11), the above-mentioned formula a(Dcr
+ -Dav) The density value [)o is determined according to 10Dcn, and the density value DO is output.

In the performance (iii), the input C degree Dcn is replaced with the average density Qav' of surrounding pixels including the pixel of interest, and this is output as the output density value DO.

The multiplexer 8 selects one of these three output density values according to the judgment signal SL input from the judgment circuit 6 and inputs the selected value to the screen generation section 9 .

The screen generation section 9 includes a memory for storing a dither matrix consisting of nxng Idlls, a comparison section, etc. The comparison section compares the stored data of the dither matrix with the input density data from the multiplexer 8, and determines the magnitude. The comparison result consisting of binary values is output to the image output device 10. The image output device 10 reproduces the original image by, for example, turning on and off a laser beam according to the binary density signal input from the screen generation section 9.

In the image processing device 3, these 98 operations are repeatedly executed for all pixels in the image memory 2.

By performing the above processing, it is possible to faithfully reproduce each document containing different images such as halftones and text, and it is also possible to sharply emphasize the edges within the halftones. In other halftone areas, noise components are removed and smooth image quality can be achieved.

[Effects of the Invention] As explained above, according to the present invention, character images and halftone images are distinguished from each other based on the ratio of the gray level of the pixel of interest to the average value of the gray levels of surrounding pixels. It becomes possible to properly identify blurred characters or low-density characters included in a character area as characters, regardless of the content and the performance of the reading system.

Further, in the present invention, the obtained ratio is compared with two different threshold values, and based on the comparison result, the pixel of interest is in the character image area, the outline of the halftone image area, and the halftone image area other than the outline. Since it is determined which of these images it is included in, and contour enhancement processing and smoothing processing are performed according to the judgment result, blur correction and noise removal are also performed on halftone images, and character images and halftone images are The image quality of both images can be improved.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an example of a configuration when implementing the image processing method according to the present invention, FIG. 2 is a diagram showing an example of a pixel matrix, and FIG. 3 is an example of filter coefficients for unsharp signals. , FIG. 4 is a graph for explaining the effects of the present invention, FIG. 5 is a graph showing changes in sharp/unsharp signals for halftone images and character images, and FIG. 6 is a graph for explaining the effects of the present invention. This is a graph for explaining the effect of . 1... Image input device, 2... Image memory, 3...
image! 133! I! Device, 4... filter processing circuit,
5.7... Arithmetic circuit, 6... Judgment circuit, 8... Multiplexer, 9... Screen generation unit, 10...
Image output device. -, Yu! -・I No. 10 No. 4

Claims (3)

[Claims] (1) Setting a pixel matrix centered on the pixel of interest, determining the ratio between the gray level of the pixel of interest and the average gray level of the pixels in the pixel matrix, and comparing the calculated ratio with a predetermined threshold; An image processing method characterized in that it is determined whether a pixel of interest is included in a character image or a halftone image based on the comparison result. (2) Set a pixel matrix centered on the pixel of interest, find the ratio between the gray level of the pixel of interest and the average gray level of the pixels in the pixel matrix, and apply the calculated ratio to two different first and second pixels. 2, and if the ratio ≧ the first threshold, it is determined to be a text area, and if the first threshold > the ratio ≧ the second
If it is the threshold value, it is determined that it is the outline of the halftone area, if the ratio < the second threshold, it is determined that it is other than the outline of the halftone area, and if the ratio ≧ the first threshold sets the output gray level of the pixel of interest to either the maximum value or the minimum value that the gray level can take, and if the first threshold > ratio ≧ second threshold, the output gray level of the pixel of interest is subjected to a predetermined calculation. An image processing method characterized in that the output gray level is set to a value subjected to contour enhancement processing according to the formula, and when ratio<second threshold value, the output gray level is set to the average gray level of pixels in the pixel matrix. (3) The image processing method according to claim (1) or (2), wherein the average gray level is an average value of gray levels of each pixel in the pixel matrix excluding the pixel of interest.