JP5024006B2 - Image processing program and image processing apparatus - Google Patents

Description

  The present invention relates to an image processing program or the like for binarizing an input digital image, and in particular, an image processing program that can solve the problem of missing characters such as thick characters and can suppress the influence of sudden noise. Etc.

  Conventionally, binarization is performed on image data in which each pixel has multi-value density gradation values. In this process, a predetermined threshold is set, and pixels having a density gradation value larger than the threshold are converted to white pixels, and pixels having a density gradation value smaller than the threshold are converted to black pixels. Such binarization processing of an image is used when an image is read by OCR, when a form or the like is discriminated, and when a data amount is reduced and stored.

In the following Patent Document 1, in view of the problems in the conventional method of determining black and white compared to a certain threshold value, the object is to obtain binary data with good quality without any character crushing or fading. A valuation device or the like has been proposed. Specifically, the maximum gradation value in a predetermined area centered on the target pixel is obtained, and when the difference between the maximum gradation value and the target pixel value is smaller than the threshold value, white is indicated, and when the difference is larger than the threshold value, black is indicated. Has been.
JP 7-325916 A

  However, in the above-mentioned Patent Document 1, in the case of an image in which a wide range is filled in with black, the portion in the image is converted into a white pixel because the difference between the target pixel and the peripheral pixel is small, so-called The problem of missing the image occurs. On the other hand, it has been proposed to combine with an image binarized by a predetermined threshold value, but if the widely painted image that is desired to be left as a black character or the like was originally a bright color, In order to make the hollow portion black in the synthesized binarized image, it is necessary to increase the threshold value. As a result, there is a high possibility that an unnecessary portion will remain as black, which still has a problem.

  In addition, sudden noise may exist in the image. However, since the noise is likely to have a maximum gradation value within a predetermined area centered on the pixel of interest, the above method is susceptible to noise. Yes, there is a possibility of making a mistake.

  Therefore, an object of the present invention is an image processing program for binarizing an input digital image, which can solve the problem of missing characters such as thick characters and can suppress the influence of sudden noise. An image processing program is provided.

  In order to achieve the above object, according to one aspect of the present invention, there is provided an image processing program that causes a computer to execute a process of binarizing image data in which a pixel has multi-value gradation values. Each pixel is scanned in a direction generally away from the pixel of interest from the pixel close to the pixel of interest until at least one pixel having a gradation value greater than or equal to a certain difference is detected from the pixel of interest of In accordance with a difference determination step for determining whether or not the difference between the scanned pixel and the gradation value of the pixel of interest is equal to or greater than the certain difference, and according to the value of the pixel whose detected gradation value is equal to or larger than the certain difference , Causing the computer to execute a binarization processing step for determining a value after binarization of the target pixel.

  Furthermore, in the above invention, a preferable aspect is that the scanning of each pixel in the difference determination step is a pixel whose gradation value is equal to or greater than a certain difference from the target pixel, and the gradation is higher than the gradation value of the target pixel. Until two or more pixels having a small value are detected, or two or more pixels having a gradation value greater than or equal to a certain difference from the pixel of interest and having a gradation value larger than the gradation value of the pixel of interest It is performed until it is done.

  Further, in the above invention, a preferable aspect is that the scanning of each pixel in the difference determination step is performed even if detection of a pixel whose gradation value is greater than a certain difference from the target pixel is not completed. When the pixel is scanned, or when a predetermined number of pixels are scanned, the process ends. In this case, in the binarization processing step, the binarized value of the target pixel is a predetermined value. It is determined by that.

  Further, in the above invention, one aspect is characterized in that the scanning of each pixel in the difference determining step is performed by thinning out pixels by a predetermined method in the scanning direction.

  In order to achieve the above object, according to another aspect of the present invention, an image processing apparatus that executes a process of binarizing image data in which pixels have multi-value gradation values is the pixel to be processed. Each pixel is scanned in a direction that is approximately away from the pixel of interest from the pixel that is close to the pixel of interest until one or more pixels that have a certain difference in gradation value from the pixel of interest are detected. A difference determining means for determining whether or not a difference between the gradation value of the pixel and the target pixel is equal to or greater than the certain difference, and according to a value of the pixel whose detected gradation value is equal to or larger than the certain difference, And binarization processing means for determining a value after binarization of the target pixel.

  Further objects and features of the present invention will become apparent from the embodiments of the invention described below.

  Embodiments of the present invention will be described below with reference to the drawings. However, such an embodiment does not limit the technical scope of the present invention. In the drawings, the same or similar elements are denoted by the same reference numerals or reference symbols.

  FIG. 1 is a functional configuration diagram according to an embodiment of an image processing apparatus to which the present invention is applied. The image processing apparatus 2 shown in FIG. 1 is an image processing apparatus to which the present invention is applied, and the processing target pixel is detected until one or more peripheral pixels whose gradation value difference from the processing target pixel is a predetermined value or more are detected. The peripheral pixels are scanned in order from the pixel located near the pixel in the direction of moving away from the pixel to be processed, and the processing target is determined according to the value of the peripheral pixel in which the detected gradation value difference becomes a predetermined value or more. The binarization of pixels is executed to solve the above-described problem of hollowing out and to suppress the influence of sudden noise.

  The system according to the present embodiment shown in FIG. 1 includes an image reading device 1 that is a so-called scanner that reads a document of an image to be binarized, and the image processing device 2 that is provided in a host computer that is the host device. It consists of.

  The image reading device 1 reads the image of the original by an optical method according to a user operation or the like, and outputs the read image as a digital image. This digital image is image data in which each pixel has a density gradation value of each color, and when read as a color image, for example, each pixel has a density gradation value of each color of RGB (for example, 0 to 255). 256 gradations). When read in gray scale, it is expressed by a one-dimensional density gradation value (for example, 256 gradations from 0 to 255). The image reading apparatus 1 supplies such image data to the image processing apparatus 2.

  Next, the image processing device 2 is constructed in the host computer of the image reading device 1 as described above. The host computer can be constituted by a so-called personal computer and includes a CPU, a ROM, a RAM, a hard disk, an interface with other devices, etc. (not shown).

  As shown in FIG. 1, the image processing apparatus 2 functionally includes a difference determination unit 21 and a binarization processing unit 22, and provides user instructions and the like for image data supplied from the image reading apparatus 1. Based on the above, binarization processing is executed, and an output image as a result is output. Each of these units 21 to 22 includes various programs stored in the ROM of the host computer, the CPU that executes processing according to the programs, the RAM, and the like. The program corresponds to an image processing program to which the present invention is applied. For example, the program is provided as a driver program for the image reading apparatus 1. Further, the program can be provided in the host computer by installing the program stored in a storage medium such as a CD in the host computer or by downloading the program from a predetermined site to the host computer via the Internet or the like. .

  The difference determination unit 21 sequentially scans the peripheral pixels of the processing target pixel (target pixel) in order to perform binarization of each pixel by the processing of the CPU according to the above program, and the difference between the target pixel and each peripheral pixel. This is a part for comparing the density (luminance) gradation value difference with a predetermined threshold value. Although specific processing contents will be described later, the above-described scanning method of the difference determination unit 21 has the characteristics of the image processing apparatus 2. When the image data supplied from the image reading device 1 is color data, the difference determination unit 21 executes a process for converting the image data into gray data.

  Next, the binarization processing unit 22 is a part that executes binarization processing of each pixel based on the determination result of the difference determination unit 21 by the processing of the CPU according to the program. This process also has the characteristics of the image processing apparatus 2, and the specific contents thereof will be described later.

  In the image processing apparatus 2 according to the present embodiment having the configuration described above, each process is executed with the following contents.

  First, when processing target data is input from the image reading unit 1, the difference determination unit 21 determines whether the input image data is color data or gray data. Is stored in the RAM or the like as it is.

  On the other hand, when the input data is color data, conversion processing to gray data is performed. For example, when the input image data is expressed in RGB, the density gradation value Y (in other words, the gradation value of luminance) of each pixel is obtained by the following equation (1).

Y = (R + G + B) / 3 (1)
Note that R, G, and B are density gradation values of each color that each pixel has, and for example each take a value of 0 to 255. Further, in the formula (1), a mixing ratio of 1/3 is used for each of R, G, and B. If the sum of these three ratios is 1, Y is set as a ratio different from the formula (1). You may ask.

  The obtained gray data image data is stored in the RAM or the like.

  FIG. 2 is a diagram illustrating a processing target image after conversion into gray data. In the example shown on the left side of FIG. 2, a non-black but dark (low gradation value) character (ABC) is drawn on a light color (high gradation value) background, and on the right side. In the example shown, white characters (ABC) are drawn on a dark-colored background (low gradation value).

  When such image data is prepared, the difference determination unit 21 and the binarization processing unit 22 execute the following difference determination and binarization processing for each pixel of the target image data. The following processing is executed for all the pixels in order from the upper left pixel of the image data, for example.

  FIG. 3 is a flowchart illustrating the difference determination and binarization processing performed by the difference determination unit 21 and the binarization processing unit 22. Hereinafter, based on FIG. 3, the difference determination and binarization processing for one pixel will be described.

  First, when a processing target pixel is determined, the difference determination unit 21 scans a peripheral pixel of the processing target pixel (target pixel) (step S1). At this time, since the first peripheral pixel is scanned, one peripheral pixel closest to the target pixel is scanned. As will be described later, the scanning of the peripheral pixels (S1) is continuously performed a plurality of times for one target pixel, and the scanning is performed in order from the closest distance from the target pixel. Done.

  FIG. 4 is a diagram illustrating the scanning order. In FIG. 4, the pixel indicated by “0” is the target pixel. In this example, the peripheral pixels are sequentially scanned in the order of the arrows in the figure. In other words, scanning is performed from a pixel close to the target pixel in the order of the numbers indicated on the pixel.

  When the peripheral pixels are scanned in this way, the difference determination unit 21 determines whether or not the scanning exceeds the scanning limit (step 2). This scanning limit is a limiting condition for stopping the scanning of the peripheral pixels for the target pixel, and when all the pixels of the target image data are scanned and there are no target pixels in the scanning of the peripheral pixels, Alternatively, when the scanning of the predetermined number of pixels is finished and the number of the surrounding pixels is exceeded, it is determined that the scanning limit is exceeded (Yes in step S2).

  On the other hand, if not, it is determined that the scanning limit has not been reached (No in step S2), and the difference Yd of the gradation value Y between the scanned peripheral pixel and the target pixel is obtained (step S3). The calculation of the difference is performed by, for example, the following equation (2).

Yd = Y [i, j] −Y [in, j−m] (2)
Y [i, j] represents the tone value of the pixel of interest, and Y [i−n, j−m] represents the tone value of the surrounding pixels. Further, i represents the horizontal position of the image, j represents the vertical position of the image, and n and m are variables for designating the positions of surrounding pixels.

  Next, the difference determination unit 21 compares the obtained difference Yd with a predetermined threshold value. Specifically, the threshold value includes a threshold value Thb for determining the target pixel as black in binarization and a threshold value Thw for determining the target pixel as white in binarization. These threshold values can be set to, for example, Thb = −40 and Thw = 40 when the density gradation value of the processing target image data is represented by a value of 0 to 255.

  First, it is determined whether or not the obtained difference Yd is equal to or less than a threshold value Thb for determining the black (step S4). As a result, if it is not less than or equal to the threshold Thb (No in step S4), it is determined whether or not the difference Yd is greater than or equal to the threshold Thw for determining the white (step S5). As a result, if it is not equal to or greater than the threshold value Thw (No in step S5), that is, if the densities of the target pixel and the surrounding pixels do not change much, the process returns to step S1. Then, the next peripheral pixel is scanned in the above-described scanning order, and the same processing is repeated.

  On the other hand, if it is determined in step S4 that the difference Yd is equal to or less than the threshold value Thb for determining black (Yes in step S4), the binarization processing unit 22 increases the black determination count Drb by one. (Step S6). The black determination number Drb is the cumulative number of times that the difference Yd is determined to be equal to or less than the threshold Thb for the target pixel.

  Next, the binarization processing unit 22 compares the black determination number Drb with a predetermined determination number (threshold) Cb (step S7). The number of times of determination Cb is a predetermined multiple value, and is set to 3 times, for example. When the black determination number Drb is smaller than the determination number Cb (No in step S7), the process returns to step S1. Then, the next peripheral pixel is scanned in the above-described scanning order, and the same processing is repeated.

  On the other hand, if the black determination number Drb is equal to or greater than the determination number Cb (Yes in step S7), the binarization processing unit 22 sets the color after the binarization of the target pixel to black (step S8). Then, a value representing black is stored in a predetermined location such as the RAM.

  On the other hand, if it is determined in step S5 that the difference Yd is equal to or greater than the threshold value Thw for determining white (Yes in step S5), the binarization processing unit 22 increases the white determination count Drw by one. (Step S9). The white determination number Drw is the accumulated number of times that the difference Yd is determined to be greater than or equal to the threshold Thw for the target pixel.

  Next, the binarization processing unit 22 compares the white determination number Drw with a predetermined determination number (threshold value) Cw (step S10). The number of times of determination Cw is a predetermined multiple value, for example, 3 times. If the white determination number Drw is smaller than the determination number Cw (No in step S10), the process returns to step S1. Then, the next peripheral pixel is scanned in the above-described scanning order, and the same processing is repeated.

  On the other hand, when the white determination frequency Drw is equal to or greater than the determination frequency Cw (Yes in step S10), the binarization processing unit 22 sets the color after binarization of the target pixel to white (step S11). Then, a value representing white is stored in a predetermined location such as the RAM.

  On the other hand, if it is determined in step 2 described above that the scanning limit has been exceeded (Yes in step S2), the binarization processing unit 22 determines the color after binarization of the target pixel in advance. The set black or white is set (step S12). Then, a value representing the determined black or white is stored in a predetermined location such as the RAM. The black or white color set in advance is appropriately determined according to the content of the image data to be processed and the purpose of binarization.

  The process described above is executed for all the pixels of the target image, the binarization process in the image processing apparatus 2 is finished, and the binarized image is output.

  FIG. 5 is a diagram for explaining the binarization processing in the image processing apparatus 2. FIG. 5B is a processing target image illustrated on the left side of FIG. 2, and FIG. 5A is an enlarged view of an area indicated by S in the image of FIG. 5B.

  In FIG. 5A, when the above-described processing in the image processing apparatus 2 is executed with the pixel indicated by x1 as the target pixel, the surrounding pixels are scanned in the order indicated by the arrows in the drawing. Until the peripheral pixel y11 is scanned, since the density difference between the target pixel x1 and the peripheral pixel is small, the black determination number Drd and the white determination number Drw described above are zero.

  Thereafter, when the scanning reaches the peripheral pixel y11, the density difference between the target pixel x1 and the peripheral pixel y11 becomes equal to or less than the threshold Thb, and the black determination count Drd becomes 1. Thereafter, similar processing is performed on the peripheral pixels y12 and y13. In the processing on the peripheral pixel y13, the black determination number Drd becomes equal to or greater than the determination number Cb (three times), and the target pixel x1 is set as a black pixel.

  Further, when the pixel indicated by x2 is the target pixel, the same processing is executed in the order indicated by the arrows in the figure, and the density of the target pixel x1 and the peripheral pixel y11 at the peripheral pixels y21, y22, and y23. The difference is equal to or greater than the threshold value Thw, and in the processing of the peripheral pixel y23, the white determination number Drw is equal to or greater than the determination number Cw (three times), and the target pixel x2 is set as a white pixel.

  As described above, by using the image processing apparatus 2, a character portion desired to be left after binarization remains black, and an unnecessary background portion is erased after binarization.

  FIG. 6 is a diagram for explaining the effect of making the determination times Cb and Cw plural. FIG. 6 illustrates a part of the processing target image. A gray part indicated by G is an area to be left after binarization of characters and the like, and a white part indicated by W is unnecessary after binarization. This is the background part. Further, a portion indicated by z is a sudden black noise and has a darker color than the gray portion.

  In this state, when the above process is executed with x3 in the figure as the target pixel, the process for each peripheral pixel proceeds as indicated by the arrow in the figure, and at the pixel z, the difference Yd becomes the threshold value Thw. Thus, the number of white determinations Drw is 1. Thereafter, in the pixels y31, y32, and y33 located in the background, the difference Yd is equal to or less than the threshold Thb, and the black determination number Drb is 3 in the pixel y33. Therefore, at this point, the target pixel x3 is a black pixel.

  As described above, in the present image processing apparatus 2, since the final determination of binarization is not performed by a single determination based on the threshold values Thb and Thw, but is performed at a stage where the determination is performed a plurality of times. Even if noise is present, the processing is not performed only by the determination about the single pixel, and there is little risk of erroneous determination. In the example shown in FIG. 6 described above, if the final determination of the binarization is made by one determination at the time of the pixel z, the target pixel x3 becomes a white pixel, which is not preferable. As described above, in the present image processing apparatus 2, correct binarization is realized without being affected by noise.

  FIG. 7 is a diagram exemplifying processing results by the image processing apparatus 2. FIG. 7A illustrates the result of binarization processing performed on the processing target image illustrated in FIG. 2 by the image processing apparatus 2. FIG. 7B shows an example when the same processing target image is processed by a conventional apparatus.

  As can be seen from the figure, the image binarized by the image processing apparatus 2 does not have a void where the central portion of the character is missing, and the amount of what appears to be noise is reduced.

  In the above description, the threshold values Thb and Thw are different from each other. However, the absolute value of the difference Yd is compared with one threshold value Th (for example, 40), and the magnitude relationship between the gradation values of the target pixel and the surrounding pixels is compared. Alternatively, the black determination number Drb or the white determination number Drw may be incremented.

  The scanning order shown in FIG. 4 is merely an example, and any other order may be used as long as scanning starts in the vicinity of the pixel of interest and moves away from the pixel of interest. Note that the scanning order shown in FIG. 4 is not scanned in the order of the shortest distance when viewed from the straight line distance from the target pixel, but if viewed from the target pixel to the surrounding pixels, Scanning is performed in ascending order of the number, and scanning is performed in a direction that is generally away from the target pixel. It is also possible to scan in the order that the linear distance is short.

  Further, when setting the above-described peripheral pixel scanning limit (S2 in FIG. 3) by the number of scanning of the peripheral pixels, it is performed as follows. First, the number of times that the above-described scanning and difference determination processing can be performed in one second is measured in advance. Then, the number of times of measurement is multiplied by the target processing time (seconds) for the entire image data, and the value is divided by the total number of pixels of the image data. The number of times obtained as a result is set as the number of scans at the scan limit.

  For example, when the number of times that processing can be performed in one second is 70000000 (times / second), the target processing time is 5 (seconds), and the total number of pixels of the image data is 1000000, one pixel of interest In contrast, the number of scans (scan limit) of peripheral pixels that can be performed is set to 350.

  In this way, by setting the scanning limit, the binarization processing can be completed within a predetermined time, and the processing time is unnecessarily reduced even for an image with a small difference in shading and an increased number of scans. There is no need to increase it. Note that the difference in processing time due to the difference in the processing target images can be reduced by making the scanning limit frequency variable according to the number of pixels of the processing target image.

  In the above-described peripheral pixel scanning, all the pixels are scanned in order from the pixel closest to the target pixel without skipping the pixels. However, the peripheral pixels to be scanned may be thinned out in order to improve the processing speed. good. FIG. 8 is a diagram illustrating a scanning order for improving the processing speed. The example shown in FIG. 8A shows a case where scanning is performed while thinning out every other pixel, and scanning is performed in the order of the numbers assigned to the pixels from the target pixel (0). Pixels that are not numbered are not scanned.

  FIG. 8B is limited to a pixel located in the left / right / up / down direction of the pixel of interest and a pixel located in the diagonal direction of the square of the pixel of interest after exceeding a certain range M in the vicinity of the pixel of interest. Scan. Note that, within the certain range M, scanning is performed for all peripheral pixels unless scanning is stopped.

  In the above description, the processing of the difference determination unit 21 and the binarization processing unit 22 is executed based on a program. However, these processing units may be configured by hardware.

  As described above, the image processing apparatus 2 according to the present embodiment determines monochrome according to the relative relationship between the pixel to be processed and the surrounding area, instead of simply determining monochrome using a single threshold. Therefore, even when a character or the like to be left is a light color, or when a background color of a character or the like is a dark color, the character can be binarized so that it can be easily seen.

  In the present image processing apparatus 2, the pixels in the center of the character are also scanned to the surrounding pixels up to the background outside the character, and are determined to be black by the above-described determination. The problem can be solved.

  Further, in the present image processing apparatus 2, when the density difference between the target pixel and the surrounding pixels becomes larger than the predetermined threshold value once, the black and white of the target pixel is not finally determined, and such a case has occurred a plurality of times. Determine black and white at the moment. Usually, sudden noise that occurs in an image has a large density difference from the surroundings but does not exist over a wide range, so the density difference between the target pixel and the surrounding pixels is larger than a predetermined threshold only by the presence of noise. It is unlikely that this will happen multiple times. Therefore, the present image processing apparatus 2 can also achieve an effect that it is less susceptible to the adverse effects of noise.

  Note that although the above-described effect on noise cannot be obtained, the black and white of the target pixel may be determined when a case where the density difference between the target pixel and the peripheral pixel becomes larger than a predetermined threshold value occurs once. . Also in this case, the effect of preventing hollowing out can be obtained.

  In the above embodiment, the system configuration includes a host computer including the image reading device 1 and the image processing device 2. However, the image processing program and the image processing device according to the present invention are limited to the system configuration. Can be applied without.

  The protection scope of the present invention is not limited to the above-described embodiment, but covers the invention described in the claims and equivalents thereof.

It is a functional block diagram concerning the example of an embodiment of the image processing device to which the present invention is applied. It is the figure which illustrated the processing target image after converting into gray data. It is the flowchart which illustrated the difference determination and the binarization process which the difference determination part 21 and the binarization process part 22 perform. It is the figure which illustrated the order of scanning of a peripheral pixel. It is a figure for demonstrating the binarization process in this image processing apparatus. It is a figure for demonstrating the effect which makes the frequency | count of determination Cb and Cw plural. It is the figure which illustrated the processing result etc. by this image processing device. It is the figure which illustrated the scanning order for a processing speed improvement.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Image reading apparatus, 2 Image processing apparatus, 21 Difference determination part (difference determination means), 22 Binarization process part (binarization process means)

Claims (5)

An image processing program for causing a computer to execute processing for binarizing image data in which pixels have multi-value gradation values,
Each pixel in the direction of moving away from the pixel of interest approximately from the pixel close to the pixel of interest until one or more pixels having a gradation value equal to or greater than a certain difference from the pixel of interest to be processed are detected. A difference determination step for determining whether or not a difference between the scanned pixel and the target pixel is equal to or greater than the predetermined difference;
A binarization processing step for determining a value after binarization of the target pixel according to the number of pixels in which the detected gradation value is equal to or greater than a certain difference. Image processing program. In claim 1,
Scanning of each pixel in the difference determination step is performed until two or more pixels having a gradation value equal to or larger than a certain difference from the target pixel and having a gradation value smaller than the gradation value of the target pixel are detected. Alternatively, the processing is performed until two or more pixels having a gradation value greater than or equal to a certain difference from the pixel of interest and having a gradation value larger than the gradation value of the pixel of interest are detected. Processing program. In claim 1 or 2,
The scanning of each pixel in the difference determination step is performed when all the pixels of the image data are scanned even when detection of a pixel whose gradation value is greater than a certain difference from the target pixel is completed, or a predetermined number of pixels are scanned. When the pixel is scanned, the process ends. In this case, the binarization processing step determines a value after binarization of the pixel of interest as a predetermined value. program. In any one of Claim 1 thru | or 3,
The image processing program characterized in that scanning of each pixel in the difference determination step is performed by thinning out pixels by a predetermined method in the scanning direction. An image processing apparatus that executes processing for binarizing image data in which pixels have multi-value gradation values,
Each pixel in the direction of moving away from the pixel of interest approximately from the pixel close to the pixel of interest until one or more pixels having a gradation value equal to or greater than a certain difference from the pixel of interest to be processed are detected. Difference determining means for determining whether a difference between the scanned pixel and the target pixel is equal to or greater than the predetermined difference;
An image processing apparatus, comprising: binarization processing means for determining a value after binarization of the pixel of interest according to the number of pixels whose detected gradation values are equal to or larger than a certain difference.