JPH08223421A - Image processor - Google Patents

Abstract

PURPOSE: To obtain easy-to-read characters by preventing black or white level density change in a character image area. CONSTITUTION: A balck/white/edge detection section 5 detects whether or not 2-gradation binary data of some specific picture element are edge picture elements based on the data of the picture element and the adjacent picture elements. When the data are the edge picture elements, an area discrimination section 6 discriminates whether or not the picture element is at a start end of an area in an edge. A correction arithmetic section 7 corrects the data of the picture element based on correction data of surrounding picture elements to be correction data and the correction data are converted into multi-gradation binary data by a comparator 9. A selector 10 selects the 2-gradation binary data when the picture element is the edge picture element and selects multi- gradation binary data when not to obtain a final output. A correction arithmetic section 11 conducts correction arithmetic operation based on the discrimination result of correction data by the area discrimination section 6 and the final output to obtain the correction data for the picture element.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus for processing an image in which two-tone images such as characters and line drawings and multi-tone images such as photographs and paintings are mixed.

[0002]

2. Description of the Related Art Image data output devices, such as a device for exposing an image on photographic printing paper, a CRT display, and the like, are capable of halftone expression, and most printers are white. There are some that can express only two gradations of black.

There is no problem in the case of a black and white two-tone image (hereinafter referred to as "character image") such as a character or a line drawing, but a multi-tone image including halftones such as a photograph or a painting (hereinafter referred to as "photo image").
Since the latter cannot be expressed as is with a latter output device such as a printer, area modulation that changes the dot size like a photograph on a newspaper or density modulation that changes the distribution density of small dots by image processing such as dithering. The halftone expression was done by.

However, when a character image and a photographic image are mixed in an image for one page input by an image input device such as an image scanner, which pixel of interest belongs to? Unknown,
If the two-gradation processing suitable for a character image is performed on all the halftones of a photographic image, it will be lost at all. Further, if the multi-gradation processing suitable for all photographic images is performed, there is a problem in that the contrast and resolution of the character image are significantly reduced.

Therefore, as disclosed in, for example, Japanese Patent Laid-Open No. 6-54181, the peak level is detected from the image data of the local area including the pixel of interest, and the background level reference value is first determined to determine the pixel of interest. By comparing each level difference with the adjacent pixel with the background level reference value, it is determined whether or not the target pixel is an edge pixel, and the target pixel is subjected to the 2-gradation processing or multi-level gradation according to the determination result. There was a suggestion to make adjustments.

Further, as disclosed in, for example, Japanese Patent Application Laid-Open No. 1-276971, a plurality of features of a pixel of interest are extracted from image data of a local area including the pixel of interest to be processed, and a threshold value corresponding to each feature is extracted. Discriminant with each other and logically judge the discriminant results, which are suitable for character images or photographic images respectively.
There has also been a proposal to set a threshold for performing the binarization process.

[0007]

According to the former proposal, if the target pixel is an edge pixel, two-gradation processing is performed, and if it is not an edge pixel, multi-gradation processing is performed. However, the edge including the target pixel is emphasized and clearly expressed.

However, in general, in image data input by an image scanner or the like and digitized by an A / D converter, a white or black portion does not always have the highest value or 0 (minimum value) of the digital data. Due to the quantization by the A / D converter, a 1-bit error or variation is inevitable.

Therefore, if the area surrounded by the edge including the pixel of interest, which is an edge pixel, is a photographic image, it does not matter at all, but if it is a character image, it may be a black or white area. Pixels that are not edge pixels are processed with multi-gradation, so even if the edge of the image is black or white, black and white inversion pixels may occur in other pixels, and as a result, the gray color is increased from the edge toward the inside of the area. Had the problem of approaching.

Also in the latter proposal, it is to determine whether a low-contrast portion such as a blur of a character is a character image or a photographic image by extracting a plurality of features of the pixel of interest and discriminating each feature. Since the thresholds for multi-gradation processing are set for the pixels other than the pixels, the same problem as in the former proposal remains true.

Such a problem occurs even when a character image area and a photographic image area are separated from each other in a one-page document (image), but black or white is present in the photographic image area. In an extreme case where the character image of is included, the edge cannot be detected, or even if the edge is detected, the density of the character image area surrounded by it changes and the character is seen by the viewer. It had the drawback that it was difficult to read (it was difficult to understand if it was a line drawing).

The present invention has been made in view of the above points, and an object of the present invention is to prevent a change in the density of black or white in a character image area to improve the image quality and make it easier to read characters.

[0013]

In order to achieve the above object, the present invention binarizes a target pixel according to the characteristics of the target pixel in a local area including the target pixel of a multivalued image to be processed. In the image processing device, a two-gradation processing means for comparing the image data of the pixel of interest with a preset threshold value for two-tone processing to obtain binary data, and the image data of the pixel of interest has already been processed in the local region. A correction calculation unit that corrects the correction data of a plurality of peripheral pixels and outputs the correction data of the pixel of interest, and compares the correction data output by the correction calculation unit with a preset threshold value for multi-gradation processing. The multi-gradation processing means that uses binary data and the level difference between the image data of the target pixel and the image data of the adjacent pixel in the local region are compared with a preset threshold for edge determination, and the target pixel is compared. Is An edge determining unit that determines whether the pixel is a pixel, binary data obtained by the two-gradation processing unit when the pixel of interest is determined to be an edge pixel by the determining unit, and multi-level data when it is determined that the pixel is not an edge pixel. Binarized data selection means for respectively selecting the binarized data by the tone processing means and outputting as binarized data of the pixel of interest,
A correction calculation unit that calculates the correction data of the pixel of interest from the image data of the pixel of interest or the correction data thereof is provided.

Further, when the target pixel is determined to be an edge pixel by the edge determining means, the target pixel is determined to be the target pixel from the respective level differences between the image data of the target pixel and the image data of the adjacent pixel in the local area. Area determination means for determining whether the area surrounded by the edge including the edge is on the start side or the end side is provided, and the correction calculation means outputs the correction data of the pixel of interest output from the correction calculation means to the binarized data. It is preferable that the correction data is calculated according to the value of the binarized data output by the selection unit and the determination result by the region determination unit.

Alternatively, the area determining means is at the start side of the area surrounded by the edge if the pixel of interest is an edge pixel for the previous row or column, and is at the end side of the edge pixel for the next row or column. It may be a means for determining.

[0016]

In the image processing apparatus configured as described above, the 2-gradation processing means performs 2-gradation processing on the image data of the pixel of interest to obtain binary data, and the multi-gradation processing means converts the image data of the pixel of interest to The correction data corrected by using the correction data of the plurality of peripheral pixels already processed by the correction calculation means is subjected to multi-gradation processing to be binarized data.

On the other hand, the edge determination means compares each level difference between the image data of the pixel of interest and the image data of its adjacent pixel with a threshold for edge determination to determine whether or not the pixel of interest is an edge pixel. If the pixel of interest is an edge pixel, the binarized data selection means outputs the binarized data that has been subjected to two gradation processing,
If it is not an edge pixel, binarized data subjected to multi-gradation processing is selected and output.

The correction calculation means calculates, from the image data of the pixel of interest or the correction data thereof, the correction data of the pixel of interest for use in the correction calculation of the subsequent pixels. Therefore, the pixels other than the edge pixels in the area surrounded by the edges are subjected to the multi-gradation processing by the correction data obtained by correcting the image data to be blacker or whiter. It does not occur and high image quality is maintained.

Further, when the edge determination means determines that the pixel of interest is an edge pixel, the area determination means surrounds the pixel of interest with edges based on the respective level differences between the image data of the pixel of interest and the image data of its adjacent pixels. The value of the finalized binarized data output by the binarized data selecting means is determined by the correction operation means by the correction operation means by determining whether it is on the start side or the end side of the area. The correction data is calculated in accordance with the determination result obtained by the area determination unit.

Therefore, the correction data of the target pixel, which is an edge pixel, is obtained by the correction data calculation including the correction depending on whether the target pixel is on the start side or the end side of the area. Pixels other than the edge pixels in the divided area are minimally affected by the correction data of the pixels outside the area.

Alternatively, if the target pixel is an edge pixel for the previous row or the previous column, the area determining means determines that the target pixel is on the start side of the area surrounded by the edge, and if it is the edge pixel for the next row or the next column. By using the means for determining that it is on the terminal end side, it is possible to easily and clearly determine whether the target pixel, which is an edge pixel, is on the starting end side or the terminal end side.

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be specifically described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of an image processing apparatus which is an embodiment of the present invention.

The image processing apparatus 1 shown in FIG. 1 includes a RAM 2 for storing image data and the like, an image data matrix,
A register (hereinafter referred to as “image data MXRG”) 3,
A correction data matrix register (hereinafter referred to as "correction data MXRG") 4, a black / white edge detection unit 5 which is a two-gradation processing unit and an edge determination unit, an area determination unit 6 which is an area determination unit, and a correction. A correction calculation unit 7 which is a calculation unit, and a dither threshold generation unit 8 which constitutes a multi-gradation processing unit.
And a selector 9, which is a binarized data selection unit, and a correction calculation unit 11, which is a correction calculation unit.

FIGS. 2 and 3 are views showing an example of a data matrix including peripheral pixels in the local regions 14 and 16 shown by broken lines centering on the pixel of interest of the image data and correction data used in the present invention. is there. In FIG. 2, all the pixels except the pixel of interest H in the area indicated by the solid line centering on the image data H of the pixel of interest indicated by a circle, that is, the adjacent area 15, are adjacent pixels.

In FIG. 3, the area enclosed by a circle and shown by the solid line containing the correction data h of the pixel of interest is the correction data area 17 of the already processed pixels of the peripheral pixels, and the other local areas are unprocessed. Since it is a pixel of, it is blank. However, the correction data h of the pixel of interest is data input from the correction calculation unit 11 after processing, and is blank before processing.

In FIG. 2 and FIG. 3, a row and a column including H and h of a pixel of interest surrounded by a circle are set as a row of interest and a column of interest, respectively. The row immediately above the row of interest is called the previous row, the row one row below is called the next row, and the adjacent columns to the left and right of the row of interest are called the front row and the next row, respectively.

In the RAM 2 shown in FIG. 1, already digitized image data of, for example, 5 bits (31 for white, 0 for black) input from an image scanner (not shown) and a code input from the correction calculation unit 11 are added. 6 bits (+31 to -3
An image data area 2a and a correction data area 2b for storing the correction data of 2) as a bitmap are provided respectively.

The image data MXRG3 is composed of a 3 × 3 matrix register for registering image data of 5 bits (absolute value) for each pixel.
A matrix of image data 5 (FIG. 2) is registered. Actually, A ~ C, F ~ registered in the process immediately before that
After shifting each image data of H, K to M by one pixel to the left, new image data D, I, N may be input from the image data area 2a of the RAM 2 and registered in each row of the next column. .

The correction data MXRG4 is composed of a 2 × 5 matrix register for registering 1-pixel 6-bit (with a sign) correction data, and is a matrix of correction data in the correction data area 17 (FIG. 3) before processing. Line correction data a
.., e, correction data f and g are respectively registered from the left of the target line, and three pixels from the right of the target line are blank.

The correction data MXRG4 is also actually shifted by one pixel to the left from the correction data group registered in the immediately preceding process, and new correction data e is registered from the correction data area 2b to the right end of the preceding row. The portion of the pixel of interest to which the correction data h is input becomes blank by the shift from the right side.

Image data MXRG3 and correction data MXR
When both G4 and the target pixel H are in a state before being processed, the registered image data from the image data MXRG3 is corrected to the black / white / edge detection unit 5 and the image data of the target pixel H is corrected as necessary. The correction data registered in the correction data MXRG4 is output to the correction calculation unit 7.

Two thresholds, a threshold value THW for white judgment and a threshold value THB for black judgment, are set in advance in the black-and-white / edge detection section 5, and image data has 5 bits (31 for white, 0 for black). In the case of), for example, each threshold is THW = 6,
TWB = -5.

The black-and-white / edge detecting section 5 calculates the value of the pixel H of interest from its adjacent pixels B, C, D, G, and
Each level difference D by subtracting the values of I, L, M and N respectively
b, Dc ... Dn are calculated, and the respective level differences are 2
The threshold values THW and THB are compared with each other to detect whether or not the pixel of interest H is an edge pixel, and whether the binarized data of the pixel of interest H obtained by the two-gradation process is white or black, and binarizes the binarized data. In addition to outputting the data to the selector 10, if it is an edge pixel, an edge signal is output to the area determination unit 6 and the selector 10.

[0034]

## EQU1 ## Db = H-B Dc = H-C Dd = H-D Dg = H-G Di = H-I Dl = H-L Dm = H-M Dn = H-N

When the edge signal is input from the black / white / edge detection unit 5, the area determination unit 6 has a level difference greater than or equal to the threshold value THW for white determination or a threshold value T for black determination in the detection unit 5.
Depending on the arrangement or combination of adjacent pixels that are less than HB,
It is determined whether the target pixel H, which is an edge pixel, is on the start side or the end side of the area surrounded by the edge including the pixel H, and if it is on the start side, the area determination signal is output to the correction calculation unit 11. To do.

FIG. 4 is a diagram showing an example of four conditional expressions by which the monochrome / edge detection unit 5 determines that the pixel is an edge pixel and the detection or determination result by the detection unit 5 and the region determination unit 6. In FIG. 4, for example, the conditional expression {Db,
Dc, Dd} ≧ THW or ≦ THB means that the respective level differences Db, Dc, Dd with respect to the adjacent pixel B, C, D of the target pixel H.
Indicates that both are above the threshold value THW or below the threshold value THB.

If the conditional expression of the condition (1) is satisfied, the target pixel H is an edge pixel, and the target pixel H becomes 2 by the two gradation processing.
The digitized data is white or black. Further, the edge including the pixel is the edge for the preceding row, and the pixel of interest H is determined to be on the start end side of the area surrounded by the edge. It is determined that the target pixel H satisfying any one of the conditional expressions (1) to (4) shown in FIG. 4 is an edge pixel, and all other target pixels are not edge pixels.

Therefore, the level difference between three adjacent pixels arranged on a straight line is the threshold value TH.
If W (= 6) or more, the target pixel H is a white edge pixel, and if it is less than or equal to the threshold value TWB (= -5), it is a black edge pixel. If the three adjacent pixels are in the previous row or the previous column, the pixel of interest H is determined to be on the start side, and if it is in the next row or the next column, it is determined to be on the end side.

The correction calculation unit 7 inputs the image data H of the target pixel from the image data MXRG3 and the correction data a to g of the peripheral pixels included in the correction data area 17 (FIG. 3) from the correction data MXRG4. The correction data X of the pixel of interest H is corrected and calculated by the arithmetic expression as shown in Equation 2, and is output to the comparator 9 and the correction calculation unit 11, respectively.

[0040]

X = H + (a / 4 + b / 2 + c + d / 2 + e / 4 + f / 2 + g) / 4 = {(c + g) × 4 + (b + d + f) × 2 + a + e} / 16 + H

That is, the correction data X is the image of the pixel of interest obtained by taking the value average of the correction data of the already processed peripheral pixels according to the magnitude of the influence of each pixel as shown in Formula 2. It is corrected by adding it to the data H.

A matrix of various thresholds used for dither processing, which is a multi-gradation processing, is set in advance in the dither threshold generation unit 8 and the number of gradations of the multi-valued image input to the image data area 2a of the RAM 2 is set. Any one of the threshold matrixes is selected according to the pattern to be expressed by the multi-gradation processing, and the threshold value of the address corresponding to the address of the pixel of interest is output to the comparator 9.

The comparator 9 compares the correction data X input from the correction calculation unit 7 with the threshold value input from the dither threshold value generation unit 8, and depending on the result, the binary data of white or black is sent to the selector 10. Output. That is, the correction data X
Is converted into binarized data by a multi-gradation processing unit including a dither threshold generation unit 8 and a comparator 9.

The selector 10 responds to the edge signal from the black / white / edge detector 5 by binarizing the binary data by the grayscale processing input from the detector 5 and the binary data by the multitone processing input by the comparator 9. Select one of the data.
That is, if the target pixel H is an edge pixel and an edge signal is input, the selector 10 selects binarized data by two-gradation processing, and if no edge signal is input, binarization by multi-gradation processing is performed. Select the data and output it as the final output image data to the next printer (not shown),
Output to the correction calculator 11.

The correction calculation unit 11 determines whether the correction data X input from the correction calculation unit 7 is the presence or absence of a region determination signal from the region determination unit 6 and whether the selected output image data input from the selector 10 is white or black. In accordance with each, for example, a calculation is performed by a correction calculation formula as shown in Formula 3, and the obtained correction data h is output to the correction data area 2b of the RAM 2 and the address circled in the correction data MXRG4.

[0046]

(3) (A) White h = X-31 Black h = X (B) White h = h + YW Black h = h + YB

In Expression 3, (A) is an arithmetic expression for determining the correction data h from the correction data X according to whether the output image data from the selector 10 is white or black, and (B) is an area determination signal input. Is an arithmetic expression for further adjusting the correction data h according to the output image data. In this example, YW = 6 and YB = -6.

FIG. 5 is a flow chart showing an example of a calculation routine of the correction calculation section 11 based on the correction calculation formula shown in the equation 3. When the routine shown in FIG. 5 is started, first, the final output image data is judged in step 1, and if it is black, the process proceeds to step 2 to use the correction data X as the correction data h as it is, and if it is white, the step 3 Go to and fix data X
The value obtained by subtracting 31 from 31 is set as the correction data h, and the process proceeds to step 4.

In step 4, it is judged whether or not the pixel of interest is the edge pixel on the starting end side based on the presence / absence of the region judging signal input from the region judging unit 6, and if not, the process jumps to the end and the starting end side is determined. If it is an edge pixel, the process proceeds to step 5 and it is determined whether it is a white edge pixel or a black edge pixel according to the output image data. If it is a white edge pixel, step 6
Then, 6 (YW) is added to the correction data h, and if it is a black edge pixel, 6 is subtracted (addition of YB) from the correction data in step 7, and each end is reached.

According to the present invention, as described above, if the pixel of interest is an edge pixel, it is binarized by the two gradation processing,
If it is not an edge pixel, it is very similar to the conventional example in that it is binarized by multi-gradation processing. However, when it is not an edge pixel, the image data H of the pixel of interest is not subjected to the multi-gradation processing as it is, but the correction data X which is corrected by using the correction data of the peripheral pixels, which has already been processed, is subjected to the multi-gradation processing. To obtain binary data.

Therefore, when the area surrounded by the edges is a character image, black-and-white reversal pixels are not generated even if the pixels which are not the edge pixels in the area are subjected to the multi-gradation processing, and go from the edges to the inside of the area. There is no problem approaching gray according to. Further, even if the area surrounded by the edges is a photographic image, the gradation disorder due to the multi-gradation processing of the correction data does not occur.

Further, this embodiment is provided with the area judging section 6,
The correction calculation unit 11 further corrects the correction data h by ± 6 depending on whether the binarized data is white or black when the area determination signal is input, that is, when the target pixel is the edge pixel on the start side. Since it is added, the influence of the image data outside the region before the edge (upper side or left side) is suppressed to the minimum.

However, in general, the correction amount YW, Y
Since B (± 6) is not so large as compared with the value of the correction data, the above-mentioned effect can be obtained even if the area determination unit 6 is not provided and the start side edge pixel is not corrected in practice. The cost can be reduced accordingly.

If the area determining means is not provided, although not shown, the area determining section 6, the determining section 6, the black / white / edge detecting section 5, and the correction calculating section in the block diagram shown in FIG. 1 are used. The line connecting 11 and 11 may be deleted. Further, in the flow chart shown in FIG. 5, steps 4 to 7 may be deleted and step 2 or step 3 may be ended.

Furthermore, in order to simplify the explanation corresponding to claim 3, the conditional expression shown in FIG. 4 is shown only in the case of four edges for the previous row, the previous column or the next row, the next column. If it is determined that all other pixels are not edge pixels, it may be determined that the pixel of interest is an edge pixel but not an edge pixel when the actual image data is processed.

FIG. 6 is a diagram showing an example of more conditional expressions and judgment results including the conditional expressions shown in FIG. 4 than the black-and-white / edge detection section 5 judges an edge pixel. The sixteen conditional expressions shown in FIG. 6 are edge pixels by comparing each level difference between the target pixel and three or four consecutive pixels adjacent to each other with a threshold value THW or THB, respectively. Whether or not it is determined, and the conditions (a, ho, ta, wo) are the same as the conditions (1, 2, 3, 4) shown in FIG. 4, respectively, and detailed description thereof will be omitted.

Further, according to this embodiment, not only when the character pixels and the photographic image are present in different areas in the original of one page, white or black characters, or in the photographic image area, or Even when black or white characters having a white or black edge are present, an image with no change in density can be obtained without impairing the resolution.

[0058]

As described above, the image processing apparatus according to the present invention can prevent the change of the density of black or white in the character image area to improve the image quality and make the character easy to read.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an image processing apparatus that is an embodiment of the present invention.

FIG. 2 is a diagram showing an example of a data matrix of a local area centered on a pixel of interest of image data used in the present invention.

FIG. 3 is a diagram showing an example of a data matrix of a local area centered on a pixel of interest of correction data used in the present invention.

FIG. 4 is a diagram showing an example of a conditional expression for determining that a pixel of interest is an edge pixel and a determination result.

5 is a flowchart showing an example of a calculation routine of a correction calculation unit shown in FIG.

FIG. 6 is a diagram showing another example of a conditional expression for determining that a target pixel is an edge pixel and a determination result.

[Explanation of symbols]

1: Image processing device 2: RAM 2a: Image data area 2b: Correction data area 3: Image data MXRG (matrix register) 4: Correction data MXRG (matrix register) 5: Monochrome / edge detection section (2 gradation processing Means, edge determination means) 6: area determination section (area determination means) 7: correction calculation section (correction calculation section) 8: dither threshold generation section 9: comparator 10: selector (binarized data selection section) 11: correction calculation Division (correction calculation means)

Claims (3)

[Claims] 1. An image processing apparatus for binarizing a pixel of interest according to the characteristics of the pixel of interest in a local area including the pixel of interest of a multivalued image to be processed, wherein image data of the pixel of interest is preset. The two-gradation processing means that compares the threshold value for the two-gradation processing with the two-valued data to obtain binary data, and the image data of the target pixel is corrected by the correction data of a plurality of peripheral pixels already processed in the local area. Correction processing means for outputting the correction data of the pixel of interest, and the correction data output by the correction processing means is compared with a preset threshold value for multi-gradation processing to obtain binary data. And comparing each level difference between the image data of the pixel of interest and the image data of an adjacent pixel in the local region with a preset threshold for edge determination to determine whether the pixel of interest is an edge pixel. Or Edge determining means for determining the binarized data by the two-gradation processing means when the pixel of interest is determined to be an edge pixel, and the multi-gradation processing means when it is determined that the pixel is not an edge pixel. Each of the binarized data is selected, and the 2
An image processing apparatus comprising: a binarized data selection unit that outputs as binarized data; and a correction calculation unit that calculates the correction data of the pixel of interest from the image data of the pixel of interest or its correction data. . 2. The image processing apparatus according to claim 1, wherein when the target pixel is determined to be an edge pixel by the edge determination unit, the image data of the target pixel and an image of an adjacent pixel in the local region are displayed. From each level difference from the data, there is provided area determination means for determining whether the target pixel is on the start side or the end side of the area surrounded by the edge including the target pixel, and the correction calculation means is provided. The correction data of the pixel of interest output from the correction calculation means is used as means for calculating correction data according to the value of the binarized data output by the binarized data selection means and the determination result by the area determination means. An image processing device characterized by the above. 3. The image processing apparatus according to claim 2, wherein the area determination unit is located on a starting end side of an area surrounded by the edge if the target pixel is an edge pixel for a preceding row or a preceding column, Alternatively, the image processing apparatus is a unit that determines that the edge pixel for the next column is on the end side.