JP2888555B2 - Image processing device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial application field) The present invention is directed to a document image in which a character portion and a photographic portion are mixed, and the resolution of the character portion and the gradation of the photographic portion The present invention relates to an image processing apparatus that performs a binarization process while maintaining high performance.

(Prior Art) Generally, in an image processing apparatus such as a document image processing apparatus capable of handling not only code information but also image information, image information read by a reading unit such as a scanner is converted into a character or a line diagram. The image information having contrast is subjected to simple binarization using a fixed threshold, and image information having a gradation property such as a photograph is binarized by a pseudo gradation means such as a dither method. This is because when the read image information is uniformly binarized by a fixed threshold value, the character
In areas such as diagrams, image quality degradation does not occur because the resolution is preserved, but in areas such as photographs, gradation is not preserved, resulting in images with image quality degradation. If gradation processing is performed uniformly by the systematic dither method, etc., the image quality will not deteriorate because the gradation is preserved in areas such as photographs, but the resolution will decrease in the areas such as characters and diagrams. This is because image quality is deteriorated and an image is formed. That is, when a single binarization method is applied to read image information, it is impossible to binarize while simultaneously satisfying the image quality of each region having different characteristics.

On the other hand, an “error diffusion method” has been proposed as a binarization method that satisfies the gradation properties of a photographic image area and also has a higher resolution than a systematic dither method for a character or diagram area. This "error diffusion method" is based on the magazine "Proceeding of the SID Vol.17 2 S
Econd Quarter 1976 75-77, `` An Adaptive Algo
As described in `` rithm for Spatial Gray Scale '', the value obtained by multiplying the binarization error when the target pixel is binarized with a certain threshold value by a predetermined weighting coefficient is still equal to a predetermined area around the target pixel. When binarizing pixels that have not yet been binarized and binarizing pixels that have not yet been binarized, this binarization error is used as a correction value to perform binarization. is there.

FIG. 6 is a block diagram showing a configuration of an image processing apparatus using the above-described “error diffusion method”. In the figure,
410 is an input image signal, 40 is a correction means 411 for correcting image information of a target pixel, 42 is a binarization means for binarizing the corrected image information of the target pixel, and 421 is a binarized image. Signal, 43 is a binarization error calculating means for calculating a binarization error of the binarized target pixel, 431 is a binarization error signal, and 44 is a weighting coefficient for storing a weighting coefficient for calculating a weighting error. The storage means, 45 is a weight error calculation means for calculating a weight error by multiplying the binary error 431 calculated by the binary error calculation means 43 by the weight coefficient of the weight coefficient storage means 44, 451 is a weight error signal, 46 is Error storage means for storing the weight error calculated by the weight error calculation means 45, and 461 is an image correction signal. Less than,
The binarization processing using the “error diffusion method” will be described in detail.

For example, an input image signal 410 obtained by reading an image with an input device such as a scanner is subjected to correction processing by an image correction signal 461 in the correction means 40 and output as a corrected image signal 411. The binarizing means 42 supplied with the corrected image signal 411 outputs the binarized image signal 421 by comparing the corrected image signal 411 with the binarization threshold Th. That is, for example, “80 ₁₆ ” (subscript “
₁₆ "indicates a hexadecimal number).
If “1” is larger than the binarization threshold Th, “1” (black pixel) is output as the binarized image signal 421, and if “1” is smaller, “0” (white pixel) is output. Next, in the binarization error calculating means 43, the corrected image signal 411 and the binarized image signal 421 (here, when the binarized image signal 421 is “0”, “0 ₁₆ ” and “1” When calculating the difference from “FF ₁₆ ”), calculate the difference
Output as 431. In the weight error calculating means 45, the weighting coefficients A, B, C, D (where A = 7/16, B = 1/16,
(C = 5/16, D = 3/16) is calculated. Here, “*” in the weight coefficient storage means 44 indicates the position of the pixel of interest, and the weighting coefficient A,
B, C, and D are multiplied to obtain a weight error 451 of four pixels around the target pixel (pixels corresponding to the positions of the weighting factors A, B, C, and D).
Is calculated. The error storage means 46 is for storing the weight error 451 calculated by the weight error calculation means 45, and the weight error 451 for four pixels calculated by the weight error calculation means 45.
Is added to the area of e _A , e _B , e _C , and e _D for the target pixel “*” and stored. The aforementioned image correction signal 46
Reference numeral 1 denotes a correction signal for the position of “*”, which is a signal obtained by accumulating weight errors for a total of four pixels calculated by the above procedure.

As described above, the “error diffusion method” minimizes the binarization error by diffusing the binarization error generated by the binarization of the target pixel to peripheral pixels and performing error compensation. Therefore, when the input image emphasizes the gradation like a photographic image, it is possible to perform the binarization process that sufficiently satisfies the gradation.

However, in an image in which resolution is more important than gradation, such as an input image such as a character image, there is a drawback that the error compensation processing is damaged and the resolution of the character portion is deteriorated. Also, in the case of an image in which a photograph and text are mixed, a weight error is calculated even when the pixel of interest is a text near the boundary between the text portion and the photo portion, and the photo portion adjacent to the text portion is binarized. The error is reflected in the correction value at the time of the correction, so that accurate error diffusion cannot be performed at the end of the photographic portion, and the image quality deteriorates.

(Problems to be Solved by the Invention) According to the present invention, since the “error diffusion method” performs a binarization process on a document image in which a character region and a photograph region are mixed by a single method, If the input image is of a type that emphasizes the gradation, such as a photographic image, a binarization process that sufficiently satisfies the gradation cannot be performed. However, in an image in which resolution is more important than gradation, such as a character image, the error compensation processing is disturbed, and the resolution of the character portion is degraded. Removes the drawback that accurate error diffusion cannot be performed near the boundary between the image and the photo area, resulting in reduced image quality, and binarization according to the characteristics of each image even for image information that includes both text and photo areas. This process improves image quality and improves text and photo areas. Provided is an image processing apparatus capable of preventing a decrease in image quality even near a boundary with a region, and further improving processing efficiency in various types of image processing by performing processing according to image characteristics. Aim.

[Structure of the Invention] (Means for Solving the Problems) An image processing apparatus according to the present invention is an image processing apparatus for processing a target image in which a character portion and a photographic portion are mixed, wherein image information of a pixel of interest in the processing target image is provided. A binarization means for binarizing, and a binarization error calculation means for calculating a binarization error from the information binarized by the binarization means and the image information,
Weighting coefficient storage means for storing a weighting coefficient for calculating a weighting error; and a weighting error calculated by multiplying the weighting coefficient stored in the weighting coefficient storage means by the binarization error calculated by the binarization error calculating means. Weighting error calculating means for calculating a feature amount, a feature amount extracting means for extracting a feature amount representing a feature of an image from image information in a predetermined range including the target pixel, and a feature amount extracted by the feature amount extracting means. Identification means for outputting an identification signal indicating whether the pixel of interest is a text part or a photograph part; and a weight error calculated by the weight error calculation means when the identification signal output from the identification means indicates a photograph part. An error truncation means for outputting a weight error calculated by the weight error calculation means and outputting a zero weight error when the identification signal output from the identification means indicates a character portion; Error storage means for storing the weight error output from, and when the identification means output from the identification means indicates a photographic part, to select to perform correction by the weight error stored in the error storage means, Selecting means for selecting the correction so that the weight error is zero when the identification signal indicates a character portion; and adding the weight error or zero selected by the selecting means to correct the image information of the target pixel. And correction means for performing the correction.

(Function) The present invention provides an image processing apparatus for processing a target image in which a text part and a photograph part are mixed, binarizing image information of a pixel of interest in a processing target image, and processing the binarized information and the image. Calculating a binarization error from the information, storing a weight coefficient for calculating a weight error, multiplying the stored binarization coefficient by the calculated binarization error to calculate a weight error; A feature amount representing a feature of an image is extracted from image information within a predetermined range including a pixel, and an identification signal indicating whether the pixel of interest is a character portion or a photograph portion is output according to the extracted feature amount. The calculated weight error is output when the identified identification signal indicates a photo portion, and the calculated weight error is truncated to zero when the output identification signal indicates a character portion. And output the weight error When the identification signal to be output indicates a photographic part, the correction is performed by selecting the correction based on the stored weight error, and when the identification signal indicates a character part, the correction is performed by setting the weight error to zero. The image information of the target pixel is corrected by adding the selected weight error or zero.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

FIG. 2 is a principle diagram showing a binarization processing method according to the present invention. In the figure, a line buffer 1 stores image information of a processing target image, and “*” indicates a position of a pixel of interest in the line buffer 1. The input image signal 21 from the line buffer 1 is supplied to the correction means 3. The correcting means 3 corrects the image information of the pixel of interest. The corrected image signal 31 corrected by the correcting means 3 is supplied to the binarizing means 4 and the binarizing error calculating means 5. The binarizing unit 4 binarizes the corrected image information of the target pixel with a predetermined threshold Th1, and the binarized image signal 41 binarized by the binarizing unit 4 is binarized. It is output to the outside as a result of the binarization processing and is also supplied to the binarization error calculation means 5. The binarization error calculating unit 5 calculates a binarization error of the binarized pixel of interest from the corrected image signal 31 and the binarization image signal 41. The binarized error signal 51 is supplied to the weight error calculating means 7. The weighting error calculating means 7 receives the weighting coefficient stored in the weighting coefficient storing means 6 and the binary error signal 51 and receives the binary error signal 51 calculated by the binary error calculating means 5.
Is multiplied by a weight coefficient of the weight coefficient storage means 6 to calculate a weight error. The weight error signal 71 calculated by the weight error calculating means 7 is supplied to the error discarding means 8. The error truncation means 8 is an identification means described later.
The identification signal 111 from 11 is used to select whether to output the weight error signal 71 or to output zero.
It is output as 81. The selected weight error signal 81 output from the error truncation means 8 is supplied to the error storage means 9 and stored as a weight error. The weight error stored in the error storage means 9 is used as an image correction signal 91 by the selection means 12.
It is supplied to.

On the other hand, the feature amount calculating means 10 calculates a maximum density difference (ΔDmax) in a predetermined area of the line buffer 1 from the image information in the area (a portion surrounded by a thick frame). The maximum density difference signal 10 output by the feature amount calculating means 10
1 is supplied to the identification means 11. The identification means 11 compares the maximum density difference signal 101 with a predetermined threshold Th2, and outputs an identification signal 111 indicating whether the pixel of interest is a photograph or a character. The identification signal 111 output from the identification means 11 is supplied to the selection means 12. When the identification signal 111 indicates that the image is a photograph, the selection unit 12 outputs the weight error stored in the error storage unit 9 as the selected image correction signal 121 and indicates that the image is a character. In this case, for example, zero is output as the selected image correction signal 121. Then, the input image signal 21 from the line buffer 1 is corrected by the selected image correction signal 121 in the correction unit 3 and supplied to the binarization unit 4.

Next, in the above configuration, the binarization processing method of the present invention will be described in detail.

For example, an input image signal 21 obtained by reading an image with an input device such as a scanner is subjected to correction processing by a selected image correction signal 121 in the correction means 3 and is output as a corrected image signal 31. The binarizing means 4 to which the corrected image signal 31 has been supplied outputs a binarized image signal 41 by comparing the corrected image signal 31 with the binarization threshold Th1. That is, for example, “80 ₁₆ ” is used as the above-described binarization threshold Th1, and if the corrected image signal 31 is larger than the binarization threshold Th1, “1” (black pixel) is output as the binarization image signal 41. In this case, “0” (white pixel) is output. Next, in the binarization error calculating means 5, the corrected image signal 31 and the binarized image signal 41
(However, here, when the binarized image signal 41 is “0”, “0 ₁₆ ”, and when it is “1”, it is “FF ₁₆ ”). Output as signal 51. In the weight error calculating means 7, the weighting coefficients A, B, C, D (where A = 7/16, B = 1/16, C = 5/16, D = 3/16) is calculated. Here, the weight coefficient storage means 6
Indicates the position of the pixel of interest, and the binarization error of the pixel of interest is multiplied by the weighting factors A, B, C, and D to obtain four pixels around the pixel of interest (the weighting factors A, B, C, and D). The weight error 71 of the pixel corresponding to the position is calculated. Then, the weight error 71 is supplied to the error discarding means 8. The error truncation unit 8 truncates the weight error 71 and selects zero when the identification signal 111 from the later-described identification unit 11 is “1”, that is, indicates that the target pixel is a character pixel.
When the identification signal 111 is “0”, that is, the pixel of interest is a photographic pixel, the weight error 71 is output as it is as the selection weight error 81.

The storage unit 9 stores the selection weight error 81 calculated by the weight error calculation unit 7 and subjected to the truncation processing according to the characteristics of the image by the weight error truncation unit 8. That is,
If the identification means 11 identifies the photo pixel, the weight error 71 for the four pixels calculated by the weight error calculation means 7 is:
Each e _A for the target pixel "*", e _B, e _C, is cumulatively stored in each area of the e _D. On the other hand, if the identification unit 11 identifies that the pixel is a character pixel, zero is added to each of the areas e _A , e _B , e _C , and e _{D for} the target pixel “*”. This means that a binary error is not diffused in the case of a character pixel. The image correction signal 91 is a signal at the position of “*”, and is a signal obtained by accumulating weight errors for a total of four pixels calculated by the above procedure.

On the other hand, the feature amount calculating means 10 calculates the maximum density difference of the image density in the (4 × 4) window (thick frame portion) including the target pixel “*”, and outputs the maximum density difference signal 101. The identification means 11 compares the maximum density signal 101 with a preset threshold Th2. If the maximum density difference signal 101 is larger than Th2, it outputs "1" indicating a character portion as the identification signal 111. "0" indicating the part is output. The selection unit 12 determines whether to output the selected image correction signal 121 according to the identification signal 111. That is, if the identification signal 111 is “0”, that is, a photographic pixel, the selected image correction signal 121
If the identification signal is "1", that is, a character pixel, zero is output as the selected image correction signal 121.

Next, an example of an image processing apparatus to which the binarization processing method is applied will be described. Parts and signals having the same functions as those in the principle diagram shown in FIG.

FIG. 1 is a schematic configuration diagram showing an image processing apparatus according to an embodiment of the present invention. This image processing apparatus inputs image information obtained by reading a document with a reading device such as an image scanner, for example, as 8-bit digital data per pixel, and binarizes the digital data. . The line buffer 1 temporarily stores such image information and provides it to the following image processing (binarization processing).

The delay means 2 receives an image signal output from the line buffer 1 in synchronization with a predetermined clock, and outputs the selected image correction signal 121 for a predetermined time, that is, an identification signal 111 described later. This is a delay until the time is reached. The correction means 3 is composed of an adder and corrects the image information of the pixel of interest. That is, the image signal 21 delayed by the delay unit 2 and a selected image correction signal 121 from the selection circuit 12 described later are added, and a corrected image signal 31 is output. The corrected image signal 31 is binarized by being compared by the binarizing means 4 at a predetermined threshold Th1, and is output as a binarized image signal 41. On this occasion,
If the corrected image signal 31 is larger than the binarization threshold Th1, “1” (black pixel) is output as the binarized image signal 41, and if it is smaller, “0” (white pixel) is output.

Next, a binarization error (EB) 51 generated in the binarization processing is calculated. The binarization error calculation means 5 includes a subtractor,
A subtraction process is performed on the corrected image signal (CI) 31 output by the correction means 3 and the binarized image signal (BI) 41 to calculate a binarized error signal 51. That is, the binarization error (EB) is obtained as EB = CI-BI (1).

The weight error calculating means 7 is constituted by a multiplier, and generates a binarization error.
51 and the weighting factor stored in the weighting factor storage means 6 are input and multiplied, and a weighting error 71 is output. The weight coefficient storage means 6 stores the four weight coefficients (A = 7/16, B = 1 /
16, C = 5/16, D = 3/16) according to the corresponding positions of the four pixels around the target pixel.
The weight error of the four pixels is obtained as e _A = A × EB (2) e _B = B × EB (3) e _C = C × EB (4) e _D = D × EB (5) . In this case, e _B is _{e B = EB- (e A +} e C + e D) ... may be obtained as (6). Then, each weight error is stored in the corresponding position of the error storage means 9. The error storage means 9 is constituted by a line memory for two lines. The image correction signal 91 is a signal read from the position of “*” in the error storage unit 9. At the position of “*” in the error storage means 9, weight errors of four pixels already processed are stored.

On the other hand, an identification signal 111 for identifying the type of image is calculated in parallel with the above operation. The identification signal 111 determines the maximum density difference as feature information in the local area including the pixel of interest from the image information output from the line buffer 1, and based on this feature information, the image information of the local area The identification signal 111 is output by determining whether the characteristic indicates a characteristic unique to a part or a characteristic as a photograph part.

That is, as shown in FIG. 5, the characteristic amount calculating means 10 calculates the maximum image density in the (4 × 4) area including the pixel of interest (the pixel indicated by oblique lines) from the image information output from the line buffer 1. Find the value and minimum value. Next, these are subtracted, and the maximum density difference signal in the (4 × 4) area, that is, feature amount information is output. Therefore, it is necessary that the line buffer 1 be constituted by line memories for four lines. The comparator 11 compares the feature amount information with a preset threshold value Th2, and outputs “1” representing a character as an identification signal when the maximum density difference signal is larger than Th2, and outputs “0” representing a photograph when the maximum density difference signal is smaller than Th2. Is output.

The above-described feature amount information is obtained as follows.

FIG. 3 is a block diagram showing a configuration example of the feature amount calculating means 10. As shown in FIG. As shown in FIG. 5, the feature value calculating means 10 calculates the maximum value and the minimum value of the density in a (4 × 4) pixel area including the target pixel, as shown in FIG. The maximum density difference is obtained by subtracting them from each other.

First, as shown in the timing chart of FIG. 4, for example, the characteristic amount calculating means 10 outputs image information (8 bits / pixel) sequentially input in units of four pixels in the column direction from the line buffer 1 in synchronization with the clock CLK. ) Are sequentially distributed to the comparators 10b, 10c, 10d, and 10e via the selector 10a. The distribution of the image information input in units of columns to the comparators 10b, 10c, 10d, and 10e by the selector 10a is performed by the clock CLK.
Signal S from the 2-bit counter 10h that operates upon receiving
The operation is controlled by E1 and SE2.

Then, the comparators 10b, 10c, 10d, and 10e compare the image information in the column direction in units of four pixels, and obtain the maximum density (MAX terminal output) and the minimum density (MIN terminal output) in that column.

The next-stage comparators 10f, 10g are the comparators 10b, 10c, 10d, 1
The signal from 0e is input at the timing of FTR1, and the maximum value and the minimum value are further obtained from the maximum value and the minimum value respectively obtained in the column direction. By the above comparison processing, the maximum value Dmax and the minimum value Dmin of the density in the area of 4 × 4 pixels are obtained, respectively, as shown in FIG.
Output at the timing of FTR2.

The subtractor 10i calculates the maximum value D of the density thus obtained.
The maximum density difference ΔDmax ΔDmax = Dmax−Dmin (7), which is the difference between max and the minimum value Dmin, is obtained.

The comparator 11 determines the type of the image based on the characteristic amount information thus obtained, that is, the maximum density difference ΔDmax, and outputs the identification signal 111.

The selection circuit 12 determines whether or not to select the image correction signal 91 read from the error storage means 9 based on the identification signal 111. That is, if the identification signal 111 is “0”, the pixel of interest is determined to be a photographic part,
The image correction signal 91 is output as the selected image correction signal 121. If the identification signal 111 is “1”, the pixel of interest is determined to be a character portion, and “0” is selected as the selected image correction signal 121.
Is output. The correction means 3 adds the selected image correction signal 121 calculated by the above-described method and the image signal 21.
This means that the correction process is performed for a photographic pixel and a binarization process is performed by an error diffusion method, and the correction process is not performed for a character pixel and a simple binarization process is performed. .

According to the image processing apparatus configured as described above, when binarizing the target pixel, whether or not to correct the target pixel based on the binarization error of the peripheral pixels is selected according to the feature of the image information. Therefore, it is possible to suppress the deterioration of the resolution of the character image seen in the conventional "error diffusion method", and to perform the binarization process that simultaneously satisfies the resolution of the character image and the gradation of the photographic image. It is possible to do. In addition, the binarization error generated when the character pixel is simply binarized is not stored as a weight error by truncation, and therefore, the weight error at the time of the simple binarization process is determined by the peripheral pixels in the following binarization process. It is not distributed to. As a result, it is possible to prevent image quality deterioration at the boundary between the character portion and the image portion of the document image in which the character portion and the photograph portion are mixed, and to obtain a clear image.

The present invention is not limited to the above embodiment. For example, the area of the reference range for extracting the feature amount is not limited to (4 × 4), and the range may be freely changed as appropriate.

In the above-described embodiment, an example in which a photograph / character is identified in units of one pixel has been described. However, identification may be performed in units of (N × N) blocks (an integer of N ≧ 2). As a result, the processing speed can be improved, and a high-speed image processing apparatus can be realized.

Further, in the above-described embodiment, an example in which the maximum density difference ΔDmax is used as the feature amount information of the image information in order to distinguish between the photograph portion and the character portion has been described.
That is, a feature amount having different properties between a text part and a photograph part, such as a value obtained by dividing the maximum density difference by the average density, or a Laplacian value which is a second derivative of the image may be used.

Further, in the above embodiment, the case where a binary output is obtained has been described. However, it is possible to output a multi-value by setting a plurality of thresholds Th1, and it is possible to obtain a multi-value laser printer, a thermal transfer printer, etc. Gradation expression becomes possible. Further, in the present invention, the value of the characteristic amount and the determination threshold are calculated based on the image signal read by the reading unit, that is, the amount corresponding to the reflectance of the image information. The identification may be performed based on the converted value (logarithm of the reciprocal of the reflectivity) and further based on the converted signal in consideration of human visual characteristics.

[Effects of the Invention] As described in detail above, according to the present invention, even for image information in which a character area and a photographic area are mixed, the binarization processing is performed in accordance with the characteristics of each image to improve the image quality. In addition to improving the image quality, it is possible to prevent the image quality from being reduced even near the boundary between the character area and the photograph area, and to improve the processing efficiency in various types of image processing by performing processing in accordance with the characteristics of the image. It is possible to provide an image processing device that can be achieved.

[Brief description of the drawings]

1 to 5 show an embodiment of the present invention.
FIG. 1 is a block diagram showing a schematic configuration of an image processing apparatus;
FIG. 2 is a diagram for explaining the principle of the binarization process, FIG. 3 is a block diagram showing the configuration of the feature amount calculating means, FIG. 4 is a timing chart showing the operation of the feature amount calculating means, and FIG. FIG. 6 is a diagram showing the concept of a pixel area for image processing, and FIG. 6 is a diagram for explaining the principle of the conventional "error diffusion method". 1 line buffer, 2 delay means, 3 correction means, 4 binarization means, 5 binarization error calculation means, 6
... Weighting coefficient storage means 7 weight error calculating means 8
... Error truncation means (processing means), 9 ... Error storage means,
10 ... feature extraction means, 11 ... comparator (identification means), 12
... Selection circuit (selection means).

Claims (1)

(57) [Claims] 1. An image processing apparatus for processing a target image in which a text part and a photograph part are mixed, a binarizing means for binarizing image information of a pixel of interest in the processing target image, A binarization error calculating unit that calculates a binarization error from the binarized information and the image information; a weight coefficient storage unit that stores a weight coefficient for calculating a weight error; Weighting error calculating means for calculating a weighting error by multiplying the weighting coefficient stored in the binarizing error calculating means by the binarizing error calculating means, and calculating an image from image information within a predetermined range including the target pixel. A feature amount extracting unit for extracting a feature amount representing a feature; an identifying unit for outputting an identification signal indicating whether the pixel of interest is a character portion or a photograph portion according to the feature amount extracted by the feature amount extracting unit; Knowledge output from identification means The weight error calculated by the weight error calculating means is output when the different signal indicates a photographic part, and the weight calculated by the weight error calculating means when the identification signal output from the identifying means indicates a character part. An error truncation unit that outputs a weight error in which the error is truncated to zero, an error storage unit that stores a weight error output from the error truncation unit, and a case where the identification signal output from the identification unit indicates a photographic part. Selecting means to perform correction based on the weight error stored in the error storage means, and selecting to perform correction by setting the weight error to zero when the identification signal indicates a character portion; Correction means for adding the weight error or zero selected by the means to correct the image information of the pixel of interest.