JP3996564B2 - Background detection method, image processing method, program, recording medium, image processing apparatus, and image forming apparatus - Google Patents

Description

  The present invention relates to a background detection method, an image processing method, a program, a recording medium storing a program, and an image in which a histogram is created based on image information of an image to be background detected and the created histogram is used. The present invention relates to a processing apparatus and an image forming apparatus.

  Conventionally, means for creating a histogram for detecting a background area of an image is known.

  For example, based on the input image data, create a histogram showing the density and its frequency, find the maximum density value and minimum density value from the histogram, and calculate the temporary background density from the maximum frequency density and the maximum density value. And a method of detecting the background density from the temporary background density and the minimum density value has been proposed (see Patent Document 1).

  Also, based on the input image data, create a histogram showing the density and its frequency, detect the maximum frequency and the density with the maximum frequency, and remove the background density to be removed from those values Means for determining a range, comparing the maximum and minimum values of the density range with the image data to determine whether the density range is within the range, and outputting the image data or the background-removed data based on the determination result It has been proposed (see Patent Document 2).

Further, based on the input color image information, it is determined whether the color image is a character image or a photographic image. If it is determined that the character image is included, the background is removed, and if it is determined that the color image is a photographic image Has proposed a means of not removing the background (see Patent Document 3).
JP-A-1-196971 Japanese Patent Laid-Open No. 3-44268 Japanese Patent No. 2984280

  However, if a histogram is simply created and the density at which the frequency is the maximum value is determined as the background density, it may be determined that the background is not actually the background. For example, in an image such as a gradation, up to a certain place in the image may be determined as an image area, and a subsequent area that is actually an image area may be determined as a background.

  Further, with the above-described means of Patent Document 1, since the entire image data is uniquely processed with a threshold value based on the detected background density, there is a possibility that a place that is not a background is actually removed as a background. There is.

  Similarly, the means of Patent Document 2 uniquely processes the entire image data with a threshold value based on the determined background density. For example, pixels having the same density as the determined background density are included in one of the photographic images. In the case where it exists in the part, a part of the photographic image may be removed as a background.

  Similarly, in the means of Patent Document 3, the background is not removed when it is determined to be a photographic image, but when a pixel having the same density as the background exists in the character image, the background is removed. This problem has not been solved.

  The problem to be solved by the present invention is to provide a background detection method, an image processing method, a program, a recording medium storing the program, an image processing apparatus, and an image forming apparatus that can detect only the background area appropriately and accurately. is there.

  A background detection method, an image processing method, a program, a recording medium storing the program, an image processing apparatus, and an image forming apparatus according to the present invention have the following configurations as means for solving the above-described problems. is there.

  (1) The whole representing the relationship between the class value of the image information and the frequency of each class value with respect to the entire area of the image, based on the image information of the image to be detected from the background composed of pixels of a plurality of rows and columns. An image histogram is created, and for each of a plurality of partial images of the image, partial image histograms representing the relationship between the class value of the image information corresponding to each partial image and the frequency of each class value are created. A first step, a second step of obtaining a maximum value of the frequency of each of the whole image histogram and the partial image histogram, and a class value of which the frequency of each of the plurality of partial image histograms is a maximum value. A third step of calculating a difference value between each other and determining whether or not there is a ground region to be detected based on the difference value; If it is determined that there is underlying region in the third step, the frequency of the entire image histogram, characterized in that it comprises a fourth step of detecting a background area based on the class value having a maximum value.

  In this configuration, a plurality of partial image histograms are created, and it is determined whether or not a uniform background area exists in the entire image based on the difference value between the class values having the maximum frequency. . Further, since it is first determined whether or not there is a background area to be detected, the background area is detected, so erroneous background detection is reduced and only the background area is accurately detected.

  (2) The method further includes the step of re-quantizing the image information to a bit number smaller than the original quantization bit number before the first step.

  In this configuration, since the whole image histogram and the partial image histogram are created based on the re-quantized image information, the creation of the histogram is simplified and the burden required for control is reduced. Also, less memory capacity is used for the histogram. Further, the background detection determination performed based on the created histogram is simplified.

  (3) In the first step, the plurality of partial image histograms include three regions of an upper plurality of rows, a middle plurality of rows, and a lower plurality of rows in the image to be detected in the background, or a plurality of left side columns, Any of the three regions of the middle multiple columns and the right multiple columns, or the six regions of the upper multiple rows, the middle multiple rows, the lower multiple rows, the left multiple columns, the middle multiple columns, and the right multiple columns It is characterized by being created about.

  In this configuration, histograms are created for three regions of upper multiple rows, middle multiple rows, and lower multiple rows, or three regions of left multiple columns, middle multiple columns, and right multiple columns. Thus, it is determined whether or not the background exists uniformly over the entire image with a small data capacity. In addition, by creating histograms for the six regions of the upper multi-row, middle multi-row, lower multi-row, left multi-column, middle multi-column, and right multi-column, is there a uniform background in the entire image? Whether or not is determined with higher accuracy.

  (4) In the first step, the plurality of partial image histograms are generated for an upper plurality of rows and at least one of a left plurality of columns or a right plurality of columns among the images to be detected for background. Features.

  In this configuration, the partial image histogram is created for the upper plurality of rows and at least one of the left plurality of columns or the right plurality of columns. Therefore, even if the image size of the input image information and the image size of the actual image are different, the normal document image is set so as to be aligned at the upper left or upper right, so it is ensured that the partial image of the document image A histogram is created. Further, it is determined whether the background exists uniformly while further suppressing the memory capacity of the partial image histogram.

  (5) The image information includes luminance information and chromaticity information. In the first step, an overall image histogram is created for each of the luminance information and chromaticity information. In the fourth step, first, luminance information is obtained. A temporary ground area is detected on the basis of the whole image histogram, and then a ground area is detected from the temporary ground area on the basis of the whole image histogram for chromaticity information.

  In this configuration, a temporary ground area is first detected based on the luminance information, and the ground area is detected by further narrowing down the temporary ground area with chromaticity information. Therefore, the groundwater basin can be detected accurately and reliably with a small amount of calculation.

  (6) In the fourth step, in addition to the class value at which the frequency of the whole image histogram is the maximum value, the class value adjacent to the class value at which the frequency is the maximum value and the class value at which the frequency is the maximum value In the case where it is the next class value and the frequency is the third largest, the base area is detected based on the class value.

  In this configuration, in addition to the class value at which the frequency of the entire image histogram has the maximum value, the background region is detected based on the class value on both sides of the class value at which the frequency has the maximum value. Further, when the frequency of the class value having the maximum frequency and the adjacent class value is the third largest, the background area is detected based on the class value. Accordingly, the base region is detected more reliably and without leakage.

  (7) The method further includes a step of setting a region obtained by subjecting the detected background region to an isolated region removal process as a base region.

  If a pixel that should be detected as a background area is not detected as a background area due to noise, or conversely, if some pixels of the same image information as the background area exist in a photographic image, the pixel is detected as the background There is a case. In this configuration, pixels in such an isolated partial region are removed based on image information of surrounding pixels. Further, the isolated region removal process is further performed on the ground region once detected. Therefore, the pixel to be detected as the background area is appropriately detected as the background area, and the pixel that is not the background area is accurately removed from the background area.

  (8) The method further includes a step of calculating the number of pixels for each detected background area and, when the number of pixels is less than a predetermined value set in advance, performing a process of removing the area from the background area. And

  In this configuration, an area where the number of pixels of the closed area is less than a predetermined value is removed from the background area. Therefore, by setting a desired threshold value in advance, a small area that is not recognized as a background area is removed from the background area. Only the ground area is detected properly. Further, this process is further performed after the isolated area removing process is performed, so that only the base area is detected more appropriately.

  The image processing method of the present invention has the following configuration as means for solving the above-described problems.

  (9) A background removal process is performed only on the background area detected by the background detection method according to any one of (1) to (8).

  In this configuration, since the background removal processing is performed only on the background area that has been properly and accurately detected by the background detection method according to any one of (1) to (8), character images, photographic images, etc. Even when pixels having the same image information as the background area exist, only the pixels in the background area are accurately removed without removing the pixels in the character image or the photographic image.

  Moreover, the program of this invention is provided with the following structures as a means for solving the above-mentioned subject.

  (10) Based on the image information of the image to be detected from the background composed of pixels of a plurality of rows and a plurality of columns, the whole representing the relationship between the class value of the image information and the frequency of each class value for the entire area of the image An image histogram is created, and for each of a plurality of partial images of the image, partial image histograms representing the relationship between the class value of the image information corresponding to each partial image and the frequency of each class value are created. A first step, a second step of obtaining a maximum value of the frequency of each of the whole image histogram and the partial image histogram, and a class value of which the frequency of each of the plurality of partial image histograms is a maximum value. A third step of calculating a difference value between each other and determining whether or not there is a ground region to be detected based on the difference value And, when it is determined in the third step that there is a background area, causing the computer to execute a fourth step of detecting the background area based on a class value at which the frequency of the whole image histogram is maximum. Features.

  In this configuration, processing based on the re-quantized image information requires less memory capacity, simplifies the detection of background detection, reduces erroneous background detection, and appropriately corrects only the background region. The background detection method that can be detected in a short time and the background removal method that can accurately remove only the pixels in the background region are executed by the computer, so that the versatility of the background detection method and the background removal method as described above is improved.

  The recording medium of the present invention has the following configuration as means for solving the above-described problems.

  (11) The whole representing the relationship between the class value of the image information and the frequency of each class value for the entire area of the image, based on the image information of the image to be detected from the background composed of pixels of a plurality of rows and columns. An image histogram is created, and for each of a plurality of partial images of the image, partial image histograms representing the relationship between the class value of the image information corresponding to each partial image and the frequency of each class value are created. A first step, a second step of obtaining a maximum value of the frequency of each of the whole image histogram and the partial image histogram, and a class value of which the frequency of each of the plurality of partial image histograms is a maximum value. A third step of calculating a difference value between each other and determining whether or not there is a ground region to be detected based on the difference value A program for causing a computer to execute a fourth step of detecting a background area based on a class value at which the frequency of the whole image histogram is maximum when it is determined that there is a background area in the third step. It is stored in a computer in a readable manner.

  In this configuration, processing based on the re-quantized image information requires less memory capacity, simplifies the detection of background detection, reduces erroneous background detection, and appropriately corrects only the background region. A computer program that executes a background detection method that can be detected easily and a background removal method that can accurately remove only pixels in the background region is easily supplied to a computer.

  The image processing apparatus of the present invention has the following configuration as means for solving the above-described problems.

  (12) A whole image histogram representing the class value of the image information and the frequency of each class value for the entire area of the image, based on the image information of the image to be detected from the background composed of pixels of a plurality of rows and columns. And a partial image histogram representing the class value of the image information corresponding to each partial image and the frequency of each class value for each of the plurality of partial images of the image to be subjected to background detection. Histogram creation means, analysis means for obtaining a class value having a maximum frequency for each of the whole image histogram and the partial image histogram, and class values having a maximum frequency for each of the plurality of partial image histograms. And determining whether there is a ground region to be detected based on the difference value. And a detecting means for detecting a temporary ground area based on a class value at which the frequency of the whole image histogram is maximum when the determining means determines that there is a ground area, and the detected temporary ground area In addition, an isolated region removing unit that uses a region obtained by further performing an isolated region removing process as a ground region, and a ground removing unit that performs a ground removing process only on the ground region obtained by the isolated region removing unit And.

  In this configuration, a plurality of partial image histograms are created, and it is determined whether or not a uniform background area exists in the entire image based on the difference value between the class values having the maximum frequency. . Further, since it is first determined whether or not there is a background area to be detected, the background area is detected, so erroneous background detection is reduced and only the background area is accurately detected. Further, since the isolated region removal process is further performed on the temporary ground region, pixels that should be detected as the ground region are properly detected as ground regions, and pixels that are not ground regions are accurately removed from the ground region. In addition, since the background removal process is performed only on the ground area that is detected appropriately and accurately, only the ground area is accurately removed.

  (13) Re-quantization means for re-quantizing the image information used by the histogram creation means to a bit number smaller than the original quantization bit number, and based on the re-quantized image information, An image histogram and the partial image histogram are created.

  In this configuration, since the whole image histogram and the partial image histogram are created based on the re-quantized image information, the creation of the histogram is simplified and the burden required for control is reduced. Also, less memory capacity is used for the histogram. Further, the background detection determination performed based on the created histogram is simplified.

  The image forming apparatus of the present invention has the following configuration as means for solving the above-described problems.

(14) An input means for inputting image information of the image to be detected on the background, an image processing apparatus according to (12) or (13), and an output means for outputting an image processed by the image processing apparatus It is characterized by providing.

  In this configuration, a plurality of partial image histograms are created, and it is determined whether or not a uniform background area exists in the entire image based on the difference value between the class values having the maximum frequency. . Further, since it is first determined whether or not there is a background area to be detected, the background area is detected, so erroneous background detection is reduced and only the background area is accurately detected. Further, since the isolated region removal process is further performed on the temporary ground region, pixels that should be detected as the ground region are properly detected as ground regions, and pixels that are not ground regions are accurately removed from the ground region. In addition, since the background removal process is performed only on the ground area that is detected appropriately and accurately, only the ground area is accurately removed. Furthermore, the processing based on the requantized image information requires less memory capacity, and the background detection determination is simplified.

  According to the present invention, the following effects can be obtained.

  (1) It can be determined whether or not a uniform background area exists in the entire image. Further, erroneous background detection can be reduced, and only the background region can be detected accurately.

  (2) The creation of a histogram can be simplified and the burden required for control can be reduced. Also, less memory capacity is used for the histogram. Furthermore, the determination of the background detection performed based on the created histogram can be simplified.

  (3) By creating histograms for three regions, it is possible to determine whether or not a background exists uniformly in the entire image with a small data capacity. In addition, by creating histograms for the six regions, it can be determined with higher accuracy whether or not the background region exists uniformly in the entire image.

  (4) Even if the image size of the input image information and the image size of the actual image are different, the partial image histogram for the document image can be generated reliably. Further, it is possible to determine whether or not the background exists uniformly while further suppressing the memory capacity of the partial image histogram.

  (5) The groundwater basin can be detected accurately and reliably with a small amount of calculation.

  (6) The base region can be detected more reliably without leakage.

  (7) A pixel to be detected as a background area can be properly detected as a background area, and pixels that are not the background area can be accurately removed from the background area.

  (8) A small region that is not recognized as a base region can be removed from the base region, and only the base region can be detected properly. Further, when this process is further performed after the isolated area removing process is performed, only the base area can be detected more appropriately.

  (9) Even when pixels having the same image information as the background area exist in the character image or the photographic image, only the pixels in the background area are accurately removed without removing the pixels in the character image or the photographic image. Can be removed.

  (10) It is possible to simplify the determination of the background detection because the memory capacity is small, and it is possible to reduce erroneous background detection, and to detect only the background area properly, and to accurately detect only the pixels in the background area. The versatility of the base removal method that can be removed can be improved.

  (11) Since the memory capacity is small, determination of background detection can be simplified, erroneous background detection can be reduced, and only the background area pixels can be detected accurately, and only the background area pixels can be accurately detected. A program for executing the background removal method that can be removed can be easily supplied to a computer.

  (12) It can be determined whether or not a uniform background area exists in the entire image. Further, erroneous background detection can be reduced, and only the background region can be detected accurately. In addition, the pixel to be detected as the background area can be appropriately detected as the background area, and the pixels that are not the background area can be accurately removed from the background area. Furthermore, it is possible to accurately remove only the base region.

  (13) The creation of a histogram can be simplified and the burden required for control can be reduced. Also, less memory capacity is used for the histogram. Furthermore, the determination of the background detection performed based on the created histogram can be simplified.

  (14) It is possible to determine whether or not a uniform background area exists in the entire image. Further, erroneous background detection can be reduced, and only the background region can be detected accurately. Furthermore, the pixel to be detected as the background area can be appropriately detected as the background area, and pixels that are not the background area can be accurately removed from the background area. In addition, only the base region can be removed accurately. Furthermore, the memory capacity can be reduced, and the background detection determination can be simplified.

  Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of a color image forming apparatus according to an embodiment of the present invention. The color image forming apparatus 10 includes a color image input device 11, a color image processing device 20, and a color image output device 12. The image data of RGB (R: red, G: green, B: blue) analog signals input by the color image input device 11 is subjected to processing described later in the color image processing device 20 to obtain CMYK (C: cyan). , M: magenta, Y: yellow, K: black) are output to the color image output device 12 as digital color signals.

  In this embodiment, the color image input device 11 is a scanner provided with a CCD (Charge Coupled Device), reads a reflected light image from a document image as an RGB analog signal, and outputs it to the color image processing device 20. The color image processing apparatus 20 will be described later. The color image output device 12 is an electrophotographic or ink jet printer that outputs image data of a document image onto a recording sheet. The color image output device 12 may be a display device such as a display.

  The color image processing apparatus 20 includes an A / D (analog / digital) conversion unit 21. In the A / D conversion unit 21, the RGB analog signal input from the color image input device 11 is quantized into, for example, an 8-bit digital signal. The RGB digital signals output from the A / D conversion unit 21 are input to the shading correction unit 22. The shading correction unit 22 performs processing for removing various distortions generated in the illumination system, the imaging system, and the imaging system of the color image input device 11 on the input RGB digital signal. The RGB signal output from the shading correction unit 22 is input to the input tone correction unit 23.

  The input tone correction unit 23 adjusts the color balance of the RGB signal (RGB reflectance signal) from which various distortions have been removed by the shading correction unit 22 and is used in the color image processing apparatus 20 such as a density signal. The signal is converted into an easy-to-handle signal. The RGB signal output from the input tone correction unit 23 is output to the color space conversion unit 24.

  In the color space conversion unit 24, the RGB signal is converted into a color space of LC1C2 (L: luminance, C1 · C2: chromaticity), and the converted LC1C2 signal is output to the background detection processing unit 26. In the background detection processing unit 26, the input LC1C2 signal is requantized from 8 bits to 5 bits, for example. The requantized LC1C2 signal is output to the memory 25 and temporarily stored in the memory 25. Further, the background area of the image is detected based on the requantized LC1C2 signal, and the detection result is output to the memory 25 and temporarily stored in the memory 25. Details will be described later.

  When the detection result of the background detection processing unit 26 is stored in the memory 25, the background removal processing unit 27 stores the image data (LC1C2 signal) directly stored in the memory 25 from the color space conversion unit 24 and the background detection processing unit. 26 detection results are read from the memory 25, and background removal processing is performed on the image data (LC1C2 signal) based on the background detection results. Details will be described later. The LC1C2 signal output from the background removal processing unit 27 is input to the region separation processing unit 28.

  The region separation processing unit 28 separates each pixel in the input image into one of a character region, a halftone dot region, and a photographic region based on the LC1C2 signal. In the region separation processing unit 28, based on the separation result, a region identification signal indicating to which region each pixel belongs is sent to the spatial filter processing unit 29, the black generation and under color removal unit 32, and the gradation reproduction processing unit 34. The LC1C2 signal input from the background removal processing unit 27 is output to the spatial filter processing unit 29 as it is.

  In the spatial filter processing unit 29, the image data of the LC1C2 signal input from the region separation processing unit 28 is subjected to spatial filter processing using a digital filter based on the region identification signal. As a result, the spatial frequency characteristic is corrected, and blurring and graininess deterioration of the output image in the color image output device 12 are prevented. For example, a region separated into character regions by the region separation processing unit 28 is subjected to sharp enhancement processing, which is one of spatial filter processing, by the spatial filter processing unit 29 in order to improve the reproducibility of black characters or color characters. The high frequency component is emphasized. Further, the region separated into halftone dot regions by the region separation processing unit 28 is subjected to low-pass filter processing for removing the input halftone dot component by the spatial filter processing unit 29. The image data processed by the spatial filter processing unit 29 is output to the scaling processing unit 30.

  In the scaling processing unit 30, the enlargement calculation process or the reduction calculation process is performed on the image data on which the spatial filter process has been performed so as to obtain a predetermined magnification. The processed LC1C2 signal is output to the color correction unit 31.

  In the color correction unit 31, the input image data of the LC1C2 signal is converted into a CMY color space, and color correction is performed in accordance with the color image output device 12. The corrected CMY signal image data is output to the black generation and under color removal unit 32.

  The black generation and under color removal unit 32 performs a black generation process for generating a black (K) signal from the CMY three-color signals after color correction, and newly subtracts the K signal acquired by the black generation process from the original CMY signal. The CMY three-color signal is converted into a CMYK four-color signal. An example of black generation processing is shown below. For example, in the process of generating black by skeleton black (general process), the input / output characteristic of the skeleton curve is y = f (x), the input data is C, M, Y, and the output data is C ′. , M ′, Y ′, and K ′, and the UCR (Under Color Removal) rate is α (0 <α <1), the black generation and under color removal processing is expressed by the following Equation 1.

  The CMYK signal output from the black generation and under color removal unit 32 is input to the gradation reproduction processing unit 34 via the output gradation correction unit 33.

  Similar to the spatial filter processing unit 29, the gradation reproduction processing unit 34 performs predetermined processing on the image data of the CMYK signal based on the region identification signal input from the region separation processing unit 28. For example, a region separated into character regions by the region separation processing unit 28 has a high resolution suitable for reproducing high-frequency components in the gradation reproduction processing unit 34 in order to improve the reproducibility of black characters or color characters. A binarization process or a multi-value process is selected and applied so as to be suitable for a display. The region separated into halftone dot regions by the region separation processing unit 28 is an output for converting a signal such as a density signal into a halftone dot area ratio which is a characteristic value of the color image output device 12 in the output tone correction unit 33. After the gradation correction processing is performed, the gradation reproduction processing unit 34 performs gradation reproduction processing (halftone generation) in which the image is finally separated into pixels and each gradation is reproduced. Is done. Further, the region separated into the photographic regions in the region separation processing unit is subjected to binarization processing or multi-value processing so as to be suitable for a display that emphasizes gradation reproducibility.

  The image data subjected to each process in the color image processing apparatus 20 described above is temporarily stored in a storage unit (not shown), read out at a predetermined timing, and input to the color image output apparatus 12. The above processing is controlled by a CPU (Central Processing Unit) (not shown).

  In the color image processing apparatus 20, the background removal processing unit 27 is disposed immediately after the background detection processing unit 26. However, the present invention is not limited to this, and the region separation processing unit 28 and the spatial filter processing unit 29 are not limited thereto. It may be arranged later, and may be arranged anywhere after the background detection processing unit 26 and before the color correction unit 31.

  Next, the configurations of the background detection processing unit 26 and the background removal processing unit 27, which are features of the present invention, will be described. FIG. 2 is a block diagram showing the configuration of the background detection processing unit 26 and the background removal processing unit 27. The background detection processing unit 26 includes re-quantization means 26a, histogram creation means 26b, histogram analysis means 26c, background determination means 26d, background area detection means 26e, and isolated area removal means 26f. The image data composed of the luminance information L and the chromaticity information C1 and C2 output from the color space conversion unit 24 is input to the requantization unit 26a and stored in the memory 25. In the requantization means 26a, the image data composed of luminance information and chromaticity information is requantized from 8 bits to 5 bits, for example. The requantized image data is stored in the memory 25 and is output to the histogram creating means 26b.

  In the histogram creation means 26b, an entire image histogram for the entire image is created for each of the luminance information L and the chromaticity information C1 and C2 based on the requantized image data, and each of the partial images for each of the partial images. An image histogram is created for each of the luminance information L and the chromaticity information C1 and C2. FIG. 3 shows an example of a histogram. In the histogram for the luminance information L, the horizontal axis indicates the class value of the luminance information L, and the vertical axis indicates the frequency of each class value. In the image histogram for the chromaticity information C1, the horizontal axis indicates the class value of the chromaticity information C1, and the vertical axis indicates the frequency of each class value. Similarly, in the image histogram for the chromaticity information C2, the horizontal axis indicates the class value of the chromaticity information C2, and the vertical axis indicates the frequency of each class value. In the histogram analysis means 26c, a class value having the maximum frequency in each histogram is calculated.

  The background determination unit 26d calculates a difference value between the class values having the maximum frequency in each of the plurality of partial image histograms, and determines whether there is a background region to be detected based on the difference value. Is done. When the background determination unit 26d determines that there is a background region to be detected, the background region detection unit 26e reads the image data requantized by the requantization unit stored in the memory 25, The background area of the image is detected based on the class value of each pixel of the image data requantized by the quantization means and the class value at which the frequency of the entire image histogram is the maximum. In the isolated area removing unit 26f, the isolated area removal processing is performed on the detected pixels in the background area, and the processed data is output to the memory 25 as final background area map data and stored in the memory 25. Is done.

  The background removal processing unit 27 includes a 1DLUT (one-dimensional lookup table) creation unit 27a and a background removal unit 27b. In the 1DLUT creation means 27 a, a 1DLUT for background removal processing is created based on the class value that maximizes the frequency of the entire image histogram, and is stored in the memory 25. In the background removal means 27b, the brightness information of the background area is obtained based on the image data composed of the luminance information L and the chromaticity information C1 and C2 and the background area map data input to the memory 25 from the isolated area removal means 26f. Correction processing is performed using 1DLUT. As a result, the background area of the image is removed. The image data from which the background has been removed is output to the region separation processing unit 28.

  Next, a processing procedure in the background detection processing unit will be described. FIG. 4 is a flowchart showing a part of the processing procedure in the background detection processing unit.

  First, the image data separated into the luminance information L and the two chromaticity information C1 and C2 by the color space conversion unit 24 is input to the background detection processing unit 26 pixel by pixel (S1). Here, each of the input image data L, C1, and C2 is 8-bit data and takes a value from 0 to 255. In addition, the chromaticity information C1 and C2 originally have values of −128 to +127, but here, 128 is added to C1 and C2 so that they can be handled without codes, and values of 0 to 255 are used.

  Next, the luminance information L of the input image data of one pixel and the two chromaticity information C1 and C2 are sequentially requantized to 5 bits (S2). The requantized image data is stored in the memory 25, and the frequency for each class value of the luminance information L and the two chromaticity information C1 and C2 is counted up (S3). Then, it is determined whether or not the processing from S1 to S3 has been completed for all the pixels of the image to be subjected to background detection (S4). If the process has not been completed for all the pixels, the process returns to S1, and the above-described process is repeated for all the pixels of the image to be detected.

  Here, the entire image histogram of the luminance information L and chromaticity information C1 and C2 is created for the entire image, and the upper 16 rows, the middle 16 rows, the lower 16 rows, the left 16 columns, and the middle 16 columns of the image. , Partial image histograms of luminance information L and chromaticity information C1 and C2 are created for the right 16 columns, respectively. Therefore, a total of 21 histograms are created in this embodiment.

  In the process of S4, when it is determined that the processes from S1 to S3 have been completed for all the pixels, the class values of the image data L, C1, and C2 at which the frequency of each created histogram has the maximum value are calculated. (S5). Then, the class values of the luminance information L having the maximum frequency in the plurality of partial image histograms are compared, the class values of the chromaticity information C1 having the maximum frequency are compared, and the frequency has the maximum value. The class values of the chromaticity information C2 are compared (S6). As these comparison results, difference values are acquired.

  Here, if there is a background area to be detected, the background area exists substantially uniformly in the entire original image, and therefore, each partial image has a background area having substantially the same number of pixels, and the character image. It is considered that the number of pixels is larger than that of the area, the halftone image area, and the photographic image area. Therefore, as described above, each partial image histogram is created and the difference value between the class values having the maximum frequency is calculated. If the difference value is equal to or less than a predetermined threshold value, each partial image has substantially the same class value. It can be determined that there are the largest number of pixels, these pixels are pixels in the background region, and the background region exists.

  Therefore, it is determined whether or not all the difference values are equal to or less than a predetermined threshold value set in advance (S7). If all the difference values are equal to or less than the threshold value, it is determined that the background area exists. If even one of the difference values exceeds the threshold value, it is determined that there is no background area, and the background detection process ends.

  If it is determined that the background area exists, the image data L, C1, and C2 requantized to 5 bits stored in the memory 25 are read out for each pixel (S8). Then, the luminance information L of the read pixel, the class value at which the frequency of the entire image histogram for the luminance information L is the maximum value, the class value adjacent to the class value at which the frequency is the maximum value, and the frequency are the maximum value When the frequency value is the third highest value and the frequency value that is next to the next value, the value is compared with each other (S9). If the difference value as the comparison result is equal to or less than each predetermined threshold value, the read pixel is provisionally determined to be a pixel in the background area. If the difference value exceeds the threshold value, it is determined that the pixel is not a pixel in the background area, and “0” is output to the memory 25 as the background area map data.

  If it is temporarily determined that the pixel is a pixel in the background area, the two chromaticity information C1 and C2 of the read pixel and the class in which the frequency of the entire image histogram for the chromaticity information C1 and C2 is the maximum value. Each value is compared (S10). If the difference values as the comparison results are each equal to or less than a predetermined threshold value, it is determined that the read pixel is a pixel in the background area, and “1” is output to the memory 25 as background area map data. If the difference value exceeds the threshold value, it is determined that the pixel is not a pixel in the background area, and “0” is output to the memory 25 as the background area map data.

  Then, it is determined whether or not the processing from S8 to S10 has been completed for all pixels (S11). If it is determined that the process has been completed for all the pixels, the process proceeds to S12. If it is determined that the processing has not been completed for all the pixels, the process returns to S8, and the image data L, C1, and C2 of the next pixel are read out. The process continues to the next row of pixels, and finally, the processes from S8 to S10 are performed on all the pixels.

  When the processing from S8 to S10 is completed for all the pixels, the isolated region removal processing is performed on the detected background region map data (S12). In this processing, for example, data in a preset mask area is replaced with a center value by a median filter or a predetermined mask area is set for the target pixel, and “1” and “0” in the mask area are set. For example, a method is used in which the number of pixels is compared and the larger value is set as the value of the target pixel by majority vote. Which method is adopted may be arbitrarily selected depending on how the final ground region detection result is handled. That is, for example, when it is necessary to erase only isolated points for each pixel, a median filter is used, and if an isolated area of a certain size is also removed, a method based on majority in the mask area may be used. The background area map data subjected to the isolated area removal process is stored in the memory 25.

  Next, the following processing is performed in order to remove the erroneously detected background region that could not be removed by the isolated region removal processing and increase the background region detection accuracy. The number of pixels in each closed area determined to be a background area is counted, and if the number of pixels is equal to or less than a predetermined threshold value, the area is determined to be a falsely detected background area, and All the values of “1” in the area are changed to “0” (S13). That is, a small area with the number of pixels equal to or less than a predetermined value is removed from the base area. Thus, the background detection processing is completed, and then the background removal processing is started.

The background removal process will be described with reference to FIG. FIG. 5 is a flowchart showing a part of the processing procedure in the background removal processing unit 27. First, a 1DLUT is created based on the class value that maximizes the frequency of the entire image histogram for the luminance information L (S21). Next, the background area map data output in the background detection process is input from the memory 25 (S22). Then, the image data L, C1, and C2 separated into the luminance information L and the two chromaticity information C1 and C2 by the color space conversion unit 24 and stored in the memory 25 are supplied to the background removal processing unit 27 for each pixel. Input (S23).

  It is determined whether the input image data for one pixel is a background area based on the background area map data (S24). If it is determined that the area is the background area, the luminance information L is output as a value corrected by 1DLUT as the luminance information L of the pixel, and the chromaticity information C1 and C2 are output by being replaced with achromatic values. Thereby, the ground is removed (S25). If it is determined in S24 that the area is not the background area, the image data L, C1, and C2 are output as they are without being processed in S25, and the process proceeds to S26.

  Then, it is determined whether or not the above processing has been completed for all the pixels (S26). If all the pixels have not been completed, the processing returns to S23, and the above processing is repeated for all the pixels. It is. If it is determined that all the pixels have been completed, the background removal processing is completed, and the resulting image data L, C1, and C2 are output to the region separation processing unit 28.

Next, the case where the background detection process and the background removal process are performed on the document image as shown in FIG. 6 will be specifically described. FIG. 6 is an explanatory diagram illustrating an example of a document image. FIG. 7 is a diagram showing image data of each pixel in a partial image area of the document image shown in FIG.
(A) shows the input 8-bit image data, FIG. 7 (b) shows the image data requantized to 5 bits, and FIG. 7 (c) shows the detected background area map data.

The 8-bit image data L, C1, and C2 (see FIG. 7A) input for each pixel are sequentially requantized to 5-bit data as shown in FIG. 7B. For example, when the luminance information L = 192 of the upper left pixel in FIG. 7A is requantized, the following processing is performed. When “192” is represented by an 8-bit binary value, “11000000” is obtained. When this is shifted 3 bits to the right, it becomes “11111000”. Then, when obtaining the logical product of “00011111” in order to obtain the value of the lower 5 bits of this value, it becomes “00011000”. When this is expressed in decimal, it is converted into 5 bits as shown in FIG. A requantized value “24” is obtained.

  The requantized image data is stored in the memory 25, and the frequency for each class value of the luminance information L and the two chromaticity information C1 and C2 is counted up. And the histogram as shown in FIG. 8 and FIG. 9 is produced by counting up the frequency for every class value about all the pixels of an image. Here, since re-quantization is performed to 5 bits, the class values of the image data L, C1, and C2 are expressed in 32 levels from 0 to 31.

  8 is an overall image histogram for the entire image area of the original image shown in FIG. 6, FIG. 8A is an overall image histogram for luminance information L, and FIG. 8B is an illustration for chromaticity information C1. 8C is an overall image histogram for the chromaticity information C2. 9 is a partial image histogram for the partial image of the original image shown in FIG. 6, FIG. 9A is a partial image histogram for the luminance information L in the upper 16 rows, and FIG. FIG. 9C is a partial image histogram for the luminance information L of the lower 16 rows, and FIG. 9C is a partial image histogram for the luminance information L of the lower 16 rows. For convenience of explanation, FIG. 9 shows only the partial image histograms for the luminance information L of the upper 16 rows, the middle 16 rows, and the lower 16 rows, but actually the upper 16 rows, the middle 16 rows, the lower 16 rows, and the left side of the image. Partial image histograms of luminance information L and chromaticity information C1 and C2 are created for the 16 columns, the middle 16 columns, and the right 16 columns, respectively. Therefore, a total of 21 histograms are created in this embodiment. When there are memory restrictions, etc., only the upper 16 rows, the middle 16 rows, the lower 16 rows, or the left 16 columns, the middle 16 columns, the right 16 columns only, or the upper 16 rows It is also possible to determine whether or not there is a background area by creating as many partial image histograms as possible, such as only two places on the left 16 rows or only two places on the upper 16 rows and the right 16 columns. It is.

  The frequency of the partial image histogram is calculated as follows, for example. For the upper partial image, the middle partial image, and the lower partial image of the image data, the start line and the end line are previously set in the register by the CPU (not shown), and the frequency of the entire image histogram immediately before the start line is first stored in the memory 25. A value obtained by subtracting the frequency of the entire image histogram immediately before the start row stored in the memory 25 from the frequency of the entire image histogram immediately after the end row is stored in the memory 25 as the frequency of the partial image histogram. It is obtained with. In the case of the left partial image, the middle partial image, and the right partial image of the image data, the start pixel position and the end pixel position are similarly set in the register in advance by the CPU, and the start pixel position is set for each row. After counting to the end pixel position, the frequency of the row is stored in the memory 25. When the start pixel position of the next row is reached, the frequency of the previous row stored in the memory 25 is read, and the value counted from the start pixel position is similarly added. The frequency of the partial image histogram is obtained by repeating for all the rows to be counted in this way.

  Focusing only on the 9 pixels shown in FIG. 7B, the luminance information L has nine class values “24”, the chromaticity information C1 has six class values “7”, and three “18”. As for the chromaticity information C2, the class value “16” is six and the “17” is three.

Then, the class values of the image data L, C1, and C2 at which the frequency of each created histogram has the maximum value are calculated. In this case, for example, it is calculated as follows. First, the frequency of the class value “0” and the frequency of the class value “1” are compared, and the class value having the higher frequency is selected. Next, the frequency of the selected class value is compared with the frequency of the class value “2”, and the class value having the larger frequency is selected. The comparison is sequentially repeated in this way, and the finally selected class value is stored in each register as the class value having the maximum frequency. For the luminance information L of the whole image histogram, the class value having the third highest frequency is also stored in the register. In this case, for example, the frequency from the class value “0” to “2” is held first, compared with the frequency of the class value “3”, the class value having the lowest frequency is rounded down, and the remaining three class values are Retained. Then, it is compared with the frequency of the class value “4”, the same processing is repeated, and the class value having the smallest frequency among the remaining three class values is registered as the class value having the third highest frequency. Stored in

  In the case of FIG. 8, the class value “24” of the luminance information L, the class value “18” of the chromaticity information C1, and the class value “17” of the chromaticity information C2 are the class values having the maximum frequency. The class value “26” of the luminance information L is stored in the register as the class value of the luminance information L having the third highest frequency. In the case of FIG. 9, the class value “24” of the luminance information L of the upper partial image, the class value “23” of the luminance information L of the middle partial image, and the class value “24” of the luminance information L of the lower partial image are obtained. These are stored in the registers as class values having the maximum frequency.

  Next, among the partial image histograms stored in the register, the class values of the luminance information L having the maximum frequency are compared, and the class values of the chromaticity information C1 having the maximum frequency are compared. The class values of the chromaticity information C2 having the maximum frequency are compared with each other. Specifically, for example, first, the class values of the luminance information L having the maximum frequency and the classes of the chromaticity information C1 having the maximum frequency are obtained for each of the upper partial image, the middle partial image, and the lower partial image. The class values of the chromaticity information C2 having the maximum frequency are compared with each other. Next, each of the left partial image, the middle partial image, and the right partial image has class values of the luminance information L having the maximum frequency, class values of the chromaticity information C1 having the maximum frequency, and the frequency. The class values of the chromaticity information C2 that is the maximum value are compared. Finally, in each of the upper partial image and the left partial image, the class values of the luminance information L having the maximum frequency, the class values of the chromaticity information C1 having the maximum frequency, and the color having the maximum frequency The class values of the degree information C2 are compared. As these comparison results, difference values are acquired.

  If all the difference values are equal to or less than a predetermined threshold value set in advance, it is determined that a base area exists. If at least one of the difference values exceeds the threshold value, it is determined that there is no background area, and the background detection process ends. Specifically, for example, for each of the luminance information L and the chromaticity information C1 and C2 of the upper partial image, the middle partial image, and the lower partial image, the difference value between the class values having the maximum frequency is equal to or less than the threshold value 1. Yes, and for each of the luminance information L and the chromaticity information C1 and C2 of the left partial image, the middle partial image, and the right partial image, the difference value between the class values having the maximum frequency is not more than the threshold value 1. In addition, for each of the luminance information L and the chromaticity information C1 and C2 of the upper partial image and the left partial image, if the difference value between the class values at which the frequency is the maximum value is equal to or less than the threshold value 1, a ground region exists. Determined. If any one of the difference values exceeds the threshold value 1, it is determined that the background area does not exist, and the background detection process ends without outputting a signal regarding the background area. Although the threshold value is set to 1 here, the threshold value is not limited to this, and can be changed as appropriate according to the number of bits to be requantized and the adjustment of determination accuracy.

  When it is determined that the background area exists, the image data L, C1, and C2 requantized to 5 bits stored in the memory 25 are read out for each pixel. Then, the luminance information L of the read pixel, the class value at which the frequency of the entire image histogram for the luminance information L is the maximum value, the class value adjacent to the class value at which the frequency is the maximum value, and the frequency are the maximum value When the frequency value is the third highest value next to the next value and the frequency is the third largest, the value is compared with each other. If the difference value as the comparison result is equal to or less than each predetermined threshold value, the read pixel is provisionally determined to be a pixel in the background area. If the difference value exceeds the threshold value, it is determined that the pixel is not a pixel in the background area, and “0” is output to the memory 25 as the background area map data.

  Here, as described above, the background area exists substantially uniformly throughout the original image, and is considered to have more pixels than the character image area, the halftone image area, and the photographic image area. Is compared with the class value having the maximum frequency in the entire image histogram, it is possible to determine whether or not the read pixel is a background area only from the luminance information.

  In addition to the class value with the maximum frequency, the class value adjacent to the class value with the maximum frequency, and the class value adjacent to the class value with the maximum frequency, the frequency is the third. The reason for comparing with the class value when the value is larger is as follows. This is because the class value at which the frequency is the maximum value is highly likely to be a ground region existing throughout the entire image. In addition, the class values on both sides of the color image input device 11 are highly likely to be background regions that are affected by errors and density unevenness in the CCD. Further, in the case of the class value adjacent to the class value where the frequency is the maximum and the frequency is the third largest, that is, the class value “26” in FIG. This is because the class value is also likely to be a ground area.

  If it is provisionally determined that the pixel is a pixel in the background area, the two chromaticity information C1 and C2 of the read pixel and the class value at which the frequency of the entire image histogram for the chromaticity information C1 and C2 is the maximum value. Are compared with each other. If the difference values as the comparison results are each equal to or less than a predetermined threshold value, it is determined that the read pixel is a pixel in the background area, and “1” is output to the memory 25 as background area map data. If the difference value exceeds the threshold value, it is determined that the pixel is not a pixel in the background area, and “0” is output to the memory 25 as the background area map data.

  When the background determination based on the luminance information L and the background determination based on the chromaticity information C1 and C2 are completed for all the pixels, the process proceeds to an isolated area removal process described later.

  The background determination based on the luminance information L and the background determination based on the chromaticity information C1 and C2 will be specifically described by taking the case shown in FIG. 8 as an example. Here, the class value Lmax = 24 of the luminance information L having the maximum frequency, the class value L3rd = 26 of the luminance information L having the third highest frequency, and the class value C1max of the chromaticity information C1 having the maximum frequency = 18. The class value C2max = 17 of the chromaticity information C2 having the maximum frequency is set in the register. Based on these values, it is determined whether or not each pixel is a pixel in the background region using the following conditional expression.

  Expression 2 is a conditional expression of the luminance information L. In the conditional expression on the first line of Equation 2, it is determined whether or not the class value of the luminance information L of the target pixel is the class value having the maximum frequency or the class value adjacent to both. In the conditional expression on the second line of Equation 2, whether the class value of the luminance information L of the target pixel is the class value adjacent to the class value where the frequency is the third largest and the frequency is the maximum value. It is determined whether or not. If at least one of the conditional expression in the first row and the conditional expression in the second row of Expression 2 is satisfied, the background determination based on the luminance information L is provisionally determined as a pixel in the background area. In the entire image histogram of FIG. 8A, the blacked out portion is the class value that satisfies the condition of Equation 2.

  In the conditional expressions of Expression 3 and Expression 4, as in the first line of Expression 2, it is determined whether or not the frequency value of the chromaticity information C1 and C2 is the maximum value or its adjacent value. Each is judged. When a pixel satisfying the condition of Expression 2 further satisfies the conditions of Expression 3 and Expression 4, the pixel is detected as a pixel in the background area. Here, in the determination based on the luminance information L, the background area is detected under a broader condition, and in the determination based on the chromaticity information C1 and C2, the area is further limited to detect the background area.

  When verifying with the pixels shown in FIG. 7B, all nine pixels are detected under the condition of Expression 2 based on the luminance information L, but only three pixels are detected under the conditions of Expression 3 and Expression 4 based on the chromaticity information C1 and C2. It is detected as a pixel in the ground area, and the remaining pixels are removed from the ground area. As a result, background area map data as shown in FIG. 7C is detected. Here, “1” indicates a pixel in the base region, and “0” indicates a pixel that is not in the base region.

  Next, in order to improve the detection accuracy, an isolated region removal process is further performed on the detected base region map data. As a specific method for performing the isolated region removal processing, for example, a method of replacing data in a preset mask region with a median filter by a median filter, or a predetermined mask region is set for a pixel of interest, There is a method in which the number of “1” and “0” are compared and the larger value is set as the value of the target pixel by majority vote. Here, the latter method based on the majority decision in the mask region will be described with a specific example. FIG. 10A is a diagram illustrating an example of background region map data in a part of image regions, and FIG. 10B is a diagram illustrating background region map data after an isolated region removal process is performed.

  In the background area map data in FIG. 10A, for example, when attention is paid to the pixels in the second row from the top and the second column from the left, the surrounding 3 × 3 pixels are set as the mask area. Since there are seven “1” s and two “0” s in this mask region, the value of the target pixel is set to “1” by majority decision (see FIG. 10B). Then, the pixel of interest is sequentially shifted in the horizontal direction, and when it is shifted horizontally to the end of the row, the process moves to the next row and the processing is continued. At this time, there may be a case where 3 × 3 pixels cannot be set as a mask area like the pixel located at the right end or the left end. In such a case, the same processing is performed in a settable mask area (for example, 2 × 3 pixels). If “1” and “0” are the same number, for example, the value of the target pixel is not changed as it is. In this way, when the isolated removal processing is performed on all the pixels in FIG. 10A, a diagram shown in FIG. 10B is obtained. The background area map data subjected to the isolated area removal processing is stored in the memory 25.

In order to further improve the detection accuracy, the number of pixels in the area determined to be the background area is counted, and if the number of pixels is equal to or less than a predetermined threshold value, the process of removing from the background area is performed. Applied. Specifically, for example, the following processing is performed. FIG. 11 is an explanatory diagram showing a state where an area having the number of pixels equal to or smaller than a predetermined value is removed from the base area. For example, as shown in FIG. 11A, in order from the upper left pixel of the base area map data subjected to the isolated area removal processing,
The value of “1” or “0” is determined, and when “1” appears, the right, left, and lower values are viewed from that pixel, and the number of pixels having “1” in any one is counted. . Further, similarly, the number of “1” s in the closed region is sequentially counted by looking at the right, left, and lower values for the pixels in the direction in which “1” exists. If this number is less than or equal to a predetermined threshold value set in advance, the area is erroneously determined to be the background area, and all the values of “1” in the area are changed to “0”. The For example, in FIG. 11, if the threshold is set to a value of 100 pixels, an area of 100 pixels or less is regarded as inappropriate as a base area, and the values of all the pixels in the area are replaced with “0”. (Refer FIG.11 (b)). It should be determined experimentally with reference to various samples how much the threshold value is set.

  In addition to the removal of the background area that could not be removed by the isolated area removal process, this process counts the number of all the remaining “1” s, and compares it with a predetermined threshold value, thereby comparing the entire image. It can also be used to determine whether or not the background area really exists. In determining whether or not the background area exists, for example, when “1” exists only in pixels of 25% or less of the entire image, it is finally determined that the background area does not exist.

  Next, the background removal process will be described. FIG. 12 is a graph of 1DLUT. First, a 1DLUT is created based on a class value that maximizes the frequency of the entire image histogram for the luminance information L stored in the register. Specifically, first, as shown in FIG. 12, a straight line (linear with a 45 degree angle) in which input and output are linear is assumed. Next, the class value at which the frequency of the entire image histogram is the maximum value is shifted left by 3 bits to return it to 8 bits. The value obtained by adding the value of “b” to the value obtained by subtracting the value of “a” is defined as the end point of the straight line change. Then, the output value at the change end point is set to 255, and the table data is created so that the output value after the change end point is 255 by connecting with a straight line from the change start point to the change end point. The values of “a” and “b” are set based on the class value with the maximum frequency, the performance of the color image input device 11 and the color image output device 12, the background area to be removed, and the like. For example, if the value obtained by returning the class value having the maximum frequency to 8 bits is equal to or greater than a predetermined threshold, a = 10 and b = 30 are set. If the value is less than the threshold, a = 20 and b = 40. Is set. This 1DLUT is stored in the memory 25.

  Further detailed description will be made by applying specific numerical values by way of example. First, a value Lmax ′ is obtained by returning the class value Lmax at which the frequency of the entire image histogram for the luminance information L is the maximum value to an 8-bit value. By shifting right by 3 bits or multiplying by 8, it can be returned to an 8-bit value as shown in Equation 5.

  Here, for example, when Lmax ′ is a value of 200 or more, a = 10 and b = 10 are set, and when Lmax ′ is a value of 199 or less, a = 15 and b = 40 are set. Is set. Here, since Lmax ′ = 192 and 199 or less, a = 15 and b = 40 are set. When creating a 1DLUT using “a” and “b”, the following Equations 6 to 8 are used.

  FIG. 13 shows 1DLUT table data created by using the above Equations 6 to 8. The 1DLUT data stores the output value LUT (x) at the memory address (LUT start address + x) of the memory 25 so that the input value x and the output value LUT (x) correspond to each other. The

  Next, the background area map data output in the background detection process is input from the memory 25. Then, the image data L, C1, and C2 separated into the luminance information L and the two chromaticity information C1 and C2 by the color space conversion unit 24 and stored in the memory 25 are supplied to the background removal processing unit 27 for each pixel. Entered.

  It is determined whether or not the input image data for one pixel is a background area based on the background area map data. If it is determined that the area is the background area, the following processing is performed. As for the luminance information L, a value corrected by 1DLUT is output as the luminance information L of the pixel. That is, the output value LUT (x) corresponding to the input value x is output based on the table data shown in FIG. The chromaticity information C1 and C2 is output by being replaced with a value of “128” without a sign (“0” with a sign) so as to be an achromatic color. As a result, the base region is removed.

  When it is determined that it is not a background area based on the background area map data, the input image data L, C1, and C2 are output as they are together with the luminance information L and the chromaticity information C1 and C2. Then, the above process is repeated for all the pixels, and the background removal process ends. The resulting image data L, C1, and C2 are output to the region separation processing unit 28.

  FIG. 14 shows the result of the background removal processing for each of the luminance information L and the chromaticity information C1 and C2. FIG. 14A shows the background area map data, FIG. 14B shows the result of the background removal processing for the luminance information L, and FIG. 14C shows the background removal processing for the chromaticity information C1. FIG. 14D is a diagram showing the result of the background removal processing for the chromaticity information C2. Based on the background area map data and the 1DLUT, it is possible to confirm a state in which the background processing is corrected and the background is removed.

  According to the image processing device 20, a plurality of partial image histograms are created, and whether or not there is a uniform background area in the entire image based on the difference value between the class values at which the respective frequencies are the maximum values. Can be determined. In addition, since the background area is detected after first determining whether or not there is a background area to be detected, erroneous background detection can be reduced, and only the background area can be accurately detected. Further, since the isolated region removal process is further performed on the area determined to be the base area, the pixel to be detected as the base area can be appropriately detected as the base area, and the pixels that are not the base area can be detected. It can be accurately removed from the underlying region.

  In addition, since the area where the number of pixels of the closed area is equal to or less than the predetermined threshold is removed from the ground area, a small area that cannot be recognized as the ground area can be removed from the ground area, and only the ground area is appropriately Can be detected. Furthermore, by performing this process after the isolated area removal process is performed, it is possible to detect only the base area more appropriately. In addition, since the background removal process is performed only on the background area that has been detected appropriately and accurately, only the background area can be accurately removed. Furthermore, by performing processing based on the requantized image information L, C1, and C2, the memory capacity can be reduced, and the background detection determination can be simplified.

  The background detection processing and background removal processing of the above-described embodiment may be configured as a program that can be executed by a computer, stored in a recording medium (not shown), and controlled by the CPU. The recording medium may be a ROM, a RAM, a built-in hard disk, or the like fixedly provided in the computer main body, or a recording medium configured to be separable from the computer main body. Recording media configured to be separable from the computer main body include, for example, tape systems such as magnetic tapes and cassette tapes, magnetic disks such as floppy (registered trademark) disks and external hard disks, CD-ROMs, MOs, MDs, and DVDs. It may be a disk system such as an optical disk such as an IC card (including a memory card) or a card system such as an optical card. Alternatively, it may be a medium carrying a fixed program including a semiconductor memory such as a mask ROM, an EPROM (Erasable Programmable Read Only Memory), an EEPROM (Electrically Erasable Programmable Read Only Memory), or a flash ROM. Accordingly, it is possible to provide a portable recording medium on which a program for performing background detection processing and background removal processing is recorded.

  Further, the stored program may be configured to be accessed and executed by the microprocessor, or the program is read out, and the read program is installed in a program storage area (not shown) of the computer, and the program is read. May be configured to be executed.

  Furthermore, in this embodiment, it is also possible to configure a system that can connect a communication network including the Internet (not shown), and a medium that fluidly carries the program so as to download the program from the communication network. Also good. When the program is downloaded from the communication network in this way, the download program may be stored in the computer main body in advance, or may be installed from another recording medium.

  The above-described program is configured not to execute only the background detection processing and background removal processing of this embodiment, but to be executed comprehensively including other processing in the image processing apparatus 20 shown in FIG. It may be.

  As the color image input device 11, for example, a flat bed scanner, a film scanner, a digital camera, or the like is used. As the color image output device 12, for example, an image display device such as a CRT display or a liquid crystal display, or an electrophotographic or inkjet printer that outputs processing results to paper or the like is used. Further, the image forming apparatus 10 may include a modem as a communication unit for connecting to a server or the like via a network.

  Furthermore, in the above-described embodiment, the number of bits after re-quantization is set to 5 bits. However, the number of bits is not limited to this, and the number of bits may be arbitrarily set according to the detection accuracy, the memory amount, and the like. In this case, the range of the class values of the image data L, C1, and C2 to be detected as the background area may be adjusted according to the set number of bits. In the above-described embodiment, the case where the background detection process and the background removal process are performed using the requantized image data L, C1, and C2 has been described. However, the input image data L, C1 and C2 may be used as they are for background detection processing and background removal processing.

1 is a block diagram showing a schematic configuration of a color image forming apparatus according to an embodiment of the present invention. It is a block diagram which shows the function structure of a background detection process part and a background removal process part. It is a figure which shows an example of a histogram. It is a flowchart which shows a part of process sequence in a background detection process part. 4 is a flowchart showing a part of a processing procedure in a background removal processing unit 27. It is explanatory drawing which shows an example of a manuscript image. FIG. 7 is a diagram illustrating image data of each pixel in a partial image area of the document image illustrated in FIG. 6, FIG. 6A illustrates input 8-bit image data, and FIG. 6B illustrates 5 bits. Requantized image data is shown, and FIG. 6C shows detected base area map data. 6 is an overall image histogram for the entire image area of the document image shown in FIG. 6, FIG. 8A is an overall image histogram for luminance information L, and FIG. 8B is an overall image histogram for chromaticity information C1. FIG. 8C is an overall image histogram for the chromaticity information C2. FIG. 9A is a partial image histogram for the partial image of the original image shown in FIG. 6, FIG. 9A is a partial image histogram for the luminance information L of the upper 16 rows, and FIG. 9B is the luminance information of the middle 16 rows. FIG. 9C is a partial image histogram for luminance information L in the lower 16 rows. FIG. 10A is a diagram illustrating an example of background region map data in a part of image regions, and FIG. 10B is a diagram illustrating background region map data after an isolated region removal process is performed. It is explanatory drawing which shows a mode that the area | region where a pixel number is below a predetermined value is removed from a base area | region. It is a graph of 1DLUT. Table data of 1DLUT. FIG. 14A shows the background area map data, FIG. 14B shows the result of the background removal processing for the luminance information L, and FIG. 14C shows the background removal processing for the chromaticity information C1. FIG. 14D is a diagram showing the result of the background removal processing for the chromaticity information C2.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Image forming apparatus 11 Color image input apparatus 12 Color image output apparatus 20 Image processing apparatus 26 Background detection processing part 26a Requantization means 26b Histogram creation means 26c Histogram analysis means 26d Background determination means 26e Background area detection means 26f Isolated area removal means 27 Ground removal processing unit 27b Ground removal means

Claims (14)

Based on the image information of the image to be detected for background detection composed of pixels of a plurality of rows and a plurality of columns, an overall image histogram representing the relationship between the class value of the image information and the frequency of each class value for the entire area of the image A first step of creating a partial image histogram representing a relationship between a class value of the image information corresponding to each partial image and a frequency of each class value for a plurality of partial images of the image. When,
A second step of obtaining a class value at which the frequency is a maximum value of each of the whole image histogram and the partial image histogram;
A third step of calculating a difference value between class values at which the respective frequencies of the plurality of partial image histograms are maximum values, and determining whether there is a ground region to be detected based on the difference value;
And a fourth step of detecting a background area based on a class value at which the frequency of the whole image histogram is maximum when it is determined in the third step that there is a background area. Detection method.   The background detection method according to claim 1, further comprising a step of re-quantizing the image information to a bit number smaller than an original quantization bit number before the first step.   In the first step, the plurality of partial image histograms include three regions of an upper plurality of rows, a middle plurality of rows, and a lower plurality of rows in the image to be detected in the background, or a plurality of left side columns and a plurality of middle sides. Created in one of the three regions of the column and the right multiple columns, or the six regions of the upper multiple rows, the middle multiple rows, the lower multiple rows, the left multiple columns, the middle multiple columns, and the right multiple columns The background detection method according to claim 1, wherein the background detection method is performed.   In the first step, the plurality of partial image histograms are created for an upper plurality of rows and at least one of a plurality of left columns or a plurality of right columns among the images to be detected for background. The background detection method according to claim 1 or 2.   The image information includes luminance information and chromaticity information. In the first step, an overall image histogram is created for each of the luminance information and chromaticity information. In the fourth step, first, an entire image for luminance information is created. 5. The temporary ground area is detected based on a histogram, and then the ground area is detected from the temporary ground area based on an overall image histogram for chromaticity information. The background detection method described.   In the fourth step, in addition to the class value at which the frequency of the whole image histogram is the maximum value, the class value adjacent to the class value at which the frequency is the maximum value, and the class value next to the class value at which the frequency is the maximum value 6. The background detection method according to claim 1, wherein a background region is detected based on a class value when the frequency is the third largest.   7. The background detection method according to claim 1, further comprising a step of setting a region obtained by subjecting the detected background region to isolated region removal processing as a background region.   The method further includes the step of calculating the number of pixels for each detected background area, and performing processing for removing the area from the background area when the number of pixels is less than a predetermined value set in advance. Item 8. The background detection method according to any one of Items 1 to 7.   9. An image processing method, wherein a background removal process is performed only on a background area detected by the background detection method according to claim 1. Based on the image information of the image to be detected for background detection composed of pixels of a plurality of rows and a plurality of columns, an overall image histogram representing the relationship between the class value of the image information and the frequency of each class value for the entire area of the image A first step of creating a partial image histogram representing a relationship between a class value of the image information corresponding to each partial image and a frequency of each class value for a plurality of partial images of the image. When,
A second step of obtaining a class value at which the frequency is a maximum value of each of the whole image histogram and the partial image histogram;
A third step of calculating a difference value between class values at which the respective frequencies of the plurality of partial image histograms are maximum values, and determining whether there is a ground region to be detected based on the difference value;
When it is determined in the third step that there is a background area, the computer executes a fourth step of detecting the background area based on a class value at which the frequency of the whole image histogram is maximum. Program. Based on the image information of the image to be detected for background detection composed of pixels of a plurality of rows and a plurality of columns, an overall image histogram representing the relationship between the class value of the image information and the frequency of each class value for the entire area of the image A first step of creating a partial image histogram representing a relationship between a class value of the image information corresponding to each partial image and a frequency of each class value for a plurality of partial images of the image. When,
A second step of obtaining a class value at which the frequency is a maximum value of each of the whole image histogram and the partial image histogram;
A third step of calculating a difference value between class values at which the respective frequencies of the plurality of partial image histograms are maximum values, and determining whether there is a ground region to be detected based on the difference value;
A program for causing a computer to execute a fourth step of detecting a background area based on a class value at which the frequency of the whole image histogram is maximum when it is determined in the third step that there is a background area. A recording medium characterized by being stored in a readable manner. Based on the image information of the image to be detected for background detection composed of pixels of a plurality of rows and a plurality of columns, a whole image histogram representing the class value of the image information and the frequency of each class value is created for the entire area of the image. In addition, for a plurality of partial images of the image to be subjected to background detection, histogram generating means for generating partial image histograms each representing a class value of the image information corresponding to each partial image and a frequency of each class value When,
Each of the whole image histogram and the partial image histogram, an analysis means for obtaining a class value having a maximum frequency,
A determination means for calculating a difference value between class values at which the respective frequencies of the plurality of partial image histograms are maximum values, and determining whether or not there is a ground region to be detected based on the difference value;
Detecting means for detecting a temporary background area based on a class value at which the frequency of the whole image histogram is maximum when the determination means determines that there is a background area;
An isolated region removing unit that uses a region obtained by further performing an isolated region removal process on the detected temporary ground region as a ground region;
An image processing apparatus comprising: a background removal unit that performs background removal processing only on the background region obtained by the isolated region removal unit.   Re-quantization means for re-quantizing the image information used in the histogram creation means to a bit number smaller than the original quantization bit number, and based on the re-quantized image information, the whole image histogram and The image processing apparatus according to claim 12, wherein the partial image histogram is created. Input means for inputting image information of the image to be detected on the background;
An image processing apparatus according to claim 12 or 13,
An image forming apparatus comprising: an output unit configured to output an image processed by the image processing apparatus;

Images