JP2011142471A - Image processing apparatus, control method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To obtain high accuracy with simple processing concerning, especially, representative color determination in color reduction processing by tile unit in a color image including a halftone dot area. <P>SOLUTION: An image processing apparatus includes: a tile dividing means for dividing input image data into a regulated size of tiles; a color frequency distribution generating means for generating the color frequency distributions of pixels included in the divided tiles by each divided tile; a position information imparting means for imparting position information of the pixels in the tiles to each class of the generated color frequency distributions; a representative color selecting means for selecting representative colors in the tiles; and a color replacing means for replacing the color information of the pixels of the tiles with the representative colors selected by the representative color selecting means. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an image processing apparatus, a control method, and a program. In particular, the present invention relates to a color-reduced color image processing method.

  In recent years, the digitization of information has progressed, and a system that scans with a scanner or the like for digitization and stores the data, or transmits the electronic data to another apparatus has been widespread, instead of storing a paper document as it is. Also, the color used for the document data itself has been changed from black and white to color (multi-value). Electronic document data is required to have a clear character portion and a uniform color in the background portion in order to achieve improved readability and high compressibility.

  However, when a paper document is actually digitized, various noises, blurs, and unevenness occur. In particular, in a color document composed of halftone dots, unevenness and blurring due to the colors of the halftone dots themselves occur, which affects readability and image quality. For this reason, a process of selecting a representative color from the image and reducing the color is performed.

  In general, it is assumed that the color reduction processing of a scanned image is incorporated in a device such as a scanner or a multifunction peripheral and is operated in real time. Therefore, memory saving is required for the color reduction processing as well as high accuracy. For this reason, Patent Document 1 discloses a technique for dividing an input image into a plurality of blocks and performing color replacement in units of blocks. In Patent Document 2, a density histogram is calculated for each tile, and when the variance is small, it is determined as a one-color block. Otherwise, a threshold is set so that the variance ratio between the foreground color side and the foreground color side is maximized, and the foreground color and background color are separated. If each variance value is small, it is determined that the block is a two-color block, and if not, a multi-color block is determined.

International Publication No. WO2006 / 066325 JP 2004-336221 A

  When an input image is divided into tiles, there are many examples in which a part of characters or the like is torn off at the edge of the tile and exists very small in the tile. In Patent Document 1, a histogram is generated, color quantization processing is performed, the shape of the quantization area and the generation position are determined to determine whether noise or a color should be left, and a histogram is generated again to perform quantization processing. That is, since the histogram calculation with a large calculation amount is performed twice, the calculation cost becomes high.

  In addition, as shown in FIG. 3C, a histogram having character ends is highly likely to have a small variance, and is classified as one color block in Patent Document 2, and cannot leave the color of the character region to be left. The color reduction processing for each tile is a case where the representative color of the tiles is slightly different between tiles due to variations in the composition points of the halftone dots and the blurring color of the boundary of the area even if the color area is the same as viewed from the entire page. (FIG. 3B).

  In Patent Document 1, in order to reduce color unevenness between tiles, the representative colors of tiles in the vicinity of the processing target are referred to. However, since this embodiment assumes an embedded device, there is a high possibility that the color information of the tile adjacent to the tile and the background information of the entire page cannot be referred to due to memory limitations. Accordingly, an object of the present invention is to realize both high accuracy and memory saving in color reduction processing of a color image including a halftone dot region, particularly in representative color determination.

  In order to solve the above problems, the present invention has the following configuration. A tile dividing unit that divides input image data into tiles of a predetermined size, a color frequency distribution generating unit that generates a color frequency distribution of pixels included in the divided tiles for each of the divided tiles, and the generated For each class of the color frequency distribution, a position information providing unit that provides pixel position information in the tile belonging to each class, a representative color selecting unit that selects a representative color in the tile, and the representative color selecting unit Color replacement means for replacing the color information of the pixel of the tile by the selected representative color, and the representative color selection means uses the position information in the class of the color frequency distribution and the color of the pixel Necessary minimum area color determination means for determining representative color candidates from the information and the color information of the class that is the representative color candidate are integrated using the position information in the color frequency distribution class And a class color information integrating means for determining the representative color.

  The image processing apparatus of the present invention also includes a tile dividing unit that divides input image data into tiles of a predetermined size, and a color that generates a color frequency distribution of pixels included in the divided tiles for each of the divided tiles. A frequency distribution generating unit; a position information adding unit that adds position information of a pixel in the tile belonging to each class to each class of the generated color frequency distribution; the generated color frequency distribution and the adding A representative color selecting means for selecting a representative color in the tile based on the position information, and a color replacing means for replacing the color information of the pixel of the tile with the representative color selected by the representative color selecting means; Have

  According to the present invention, the present invention can achieve both high accuracy and memory saving in color-reduction processing of a color image including a halftone dot region, particularly in representative color determination. Specifically, even if the frequency is low on the histogram, it can be easily and accurately determined whether the color should be the representative color. In addition, it is possible to reduce color unevenness between tiles after color reduction.

The example figure of the apparatus configuration of the system concerning the present invention. 1 is a block diagram showing a functional configuration of an image processing apparatus according to the present embodiment. The figure for demonstrating the image data which concern on this embodiment. FIG. 3 is a block diagram showing a functional configuration of a color quantization unit according to the present embodiment. The figure of the flowchart of the histogram generation part which concerns on this embodiment. The figure which shows an example of the data structure handled by the histogram production | generation part which concerns on this embodiment. FIG. 4 is an exemplary diagram for explaining color space division according to the embodiment. The example of a figure for explaining the histogram concerning this embodiment. The figure for demonstrating the area code which concerns on this embodiment. The figure of the flowchart of the representative color selection part which concerns on this embodiment. The figure of the flowchart of one method (method3) of representative color candidate class determination concerning this embodiment. The figure of the flowchart of representative color information determination concerning this embodiment.

<First embodiment>
FIG. 1 is an apparatus configuration diagram of a system according to the first embodiment of the present invention. The image processing apparatus 100 has a configuration for realizing the present invention. The image processing apparatus 100 includes a scanner 101 that converts page information of a read document into image data, and a CPU 102 that executes a program for performing processing of the present invention on the image data. In addition, the image processing apparatus 100 inputs and outputs data to and from a work memory when executing a program, a memory 103 used for temporary storage of data, a hard disk 104 that stores programs and data, and an external device. Network I / F 105. A personal computer (hereinafter referred to as a PC) 120 is connected to the image processing apparatus 100 via a network such as a LAN 110 and receives data transmitted from the image processing apparatus 100.

  FIG. 2 is a block diagram illustrating a functional configuration of the image processing apparatus 100 according to the first embodiment. Each process shown here is realized by executing an electronic document generation program by the CPU 102, but a part or all of it may be configured and controlled by an electric circuit. The image 201 is input by the scanner 101. In this embodiment, the description will be continued assuming that an image as shown in (1) of FIG. The color quantization unit 202 performs a color reduction process to eliminate scan noise and halftone dots from the image 201. The analysis unit 203 analyzes the reduced color image and assigns attributes necessary for subsequent processing such as characters and background. The analysis result 204 is an output product of the 203 analysis unit. The analysis result 204 is then converted into an electronic file by an electronic document generation program that is continuously executed on the CPU 102 or the PC 120.

  The detailed description of the color quantization unit 202 which is the main configuration of the present invention will be continued. FIG. 4 is a block diagram showing a functional configuration of the color quantization unit 202. A tile division unit 401 divides input image data into images (tiles) of a prescribed size, and a tile 402 is an output product of the tile division unit 401. For example, the input document image data is divided into images (tiles) each having a size of 32 pixels × 32 pixels in a tile shape. The color quantization unit 202 processes the tiles 402 divided by the tile dividing unit 401 one by one. For example, when there is an input image in which the background portion indicated by (1) in FIG. 3A is a halftone dot region and characters are in the foreground, the tile dividing unit 401 uses tiles 301 to (2) in FIG. It divides like 304, and it changes to subsequent processing for every tile. The smoothing unit 403 performs a smoothing process on the tiles 402 of the input image data divided by the tile dividing unit 401 to perform smoothing. The purpose of smoothing is to reduce the influence of the color of the halftone dot composing point itself on the representative color selection, and is not an essential process in the configuration of the present invention. Therefore, any method for crushing halftone dot composing points may be used, and any of known smoothing filters can be used. In this embodiment, it is assumed that low resolution conversion is performed using an averaging filter. When the low-resolution conversion is performed, the number of pixels of the tile for which the histogram is to be generated can be reduced, resulting in an effect of reducing the load of the histogram generation process. For example, a 32 × 32 pixel size tile is smoothed by low-resolution conversion to generate a 16 × 16 pixel size smoothed tile.

  The histogram generation unit 408 receives the smoothing tile 404 as an input and generates a histogram. The histogram generation unit 408 includes a color space division unit 405, a color frequency distribution generation unit 406, and a position information addition unit 407. Details of the histogram generation unit 408 will be described later. The representative color selection unit 409 determines the number of representative colors in the tile and the color information of the representative colors from the previously generated histogram. Details of the representative color selection unit 409 will be described later. The representative color information 410 will be described later.

  The color replacement unit 411 obtains an inter-color distance between the input color information of the tile 402 in units of pixels and the representative color information 410 obtained by the representative color selection unit 409, and changes the color to the representative color closest to the input image data. Quantization processing for replacing. As a result, a quantized tile 412 is generated. The reason why the input tile before smoothing is used for color replacement is to prevent small characters from being lost or crushed. In the present invention, the Euclidean distance between the pixel of interest and the RGB space of the representative color is calculated, but the color space and the distance calculation method are not limited to this. Further, class position information that will be described in detail later may be used.

  In the representative color selection method solved by the present invention, the color information of the minute area at the edge of the tile can be selected as the representative color, but it is assumed that the color difference between the representative colors in the tile is not so large. . Therefore, by referring to the position information of the pixel to be replaced and the position information of the representative color and replacing the representative color in the area including the pixel position to be replaced, the color replacement error can be reduced. The functional configuration of the color quantization unit 202 has been described above.

  Next, details of the histogram generation unit 408 and the representative color selection unit 409, which are features of the present invention, will be described. First, details of the histogram generation unit 408 will be described. FIG. 5 is a flowchart of the histogram generation unit 408. Note that the step corresponding to the color space dividing unit 405 in FIG. 5 is S4081. Similarly, steps corresponding to the color frequency distribution generation unit 406 are S4082 to S4086, S4088, and S4089, and a step corresponding to the position information addition unit 407 is S4087.

  FIG. 6 shows an example of the data structure handled by the histogram generation unit 408. FIG. 6A shows an example of a class data structure in the histogram. The number of pixels, average color information, and position information are stored in each class. FIG. 6B shows an example of the data structure of the histogram. The histogram has the above-mentioned class arrangement structure. There are as many arrays as the number divided by the color space dividing unit 405. Hereinafter, the histogram generation unit 408 will be described with reference to FIG. In step S4081, the class level of each histogram is set. This corresponds to the function of the color space dividing unit 405. In this embodiment, a case where a general color image is assumed to be input and color space division and color frequency distribution generation are performed will be described. However, the input may be a gray image. In other words, it is clearly stated that the meanings intended by the colors described in this embodiment correspond to gray scales in addition to those corresponding to RGB values. In this embodiment, the Y / HS color space using the luminance Y in the YUV color space and the hue H in the HSV color space is divided to set the level of color information corresponding to each class of the histogram. In the present embodiment, the Y / HS color space is divided to set a plurality of classes, but the color space to be used is not limited to the Y / HS color space.

  First, the setting of the hue H level will be described. The HSV color space has a conical shape as shown in FIG. The radial direction is saturation S, and the closer to the origin of the cylinder, the closer to the achromatic color, and the farther, the chromatic color. The outer peripheral angle of the cylinder represents a hue, and indicates a color with a value of 0 to 360 degrees such as red, magenta, blue, cyan, green, yellow, and orange. The brightness V becomes darker in the direction from the upper base to the lower base. The hue direction is set to a level obtained by dividing the central portion H0 in the hue circle by the achromatic color portion and a region obtained by dividing the hue circle at an arbitrary angle by the chromatic color portions (H1 to H4).

  Next, the level setting of the luminance Y will be described. Since the luminance value Y is one-dimensional linear, it is sufficient to simply divide the luminance value of the pixel by the prescribed number of divisions. In this embodiment, the luminance direction is divided into eight. Note that the luminance Y in the YUV color space is closer to the brightness when viewed from the human eye than the brightness V in the HSV color space. For this reason, Y in the YUV color space is used for information on color brightness. Therefore, in the histogram of FIG. 8 described below, there is a relationship between the hue H and the luminance Y.

  FIG. 8 is a visualization of the histogram structure set in S4081. Such a histogram is generated for each tile, and the process proceeds to the representative color selection unit 409, which is the next process. S4082 means that the processing of S4083 to S4087 described later is repeated for all the pixels in the smoothing tile. In S4083, one pixel (hereinafter referred to as P_ds) is acquired from the smoothed tile. In S4084, P_ds color information is acquired, and from the luminance value and hue, the luminance level and hue level in the color space divided by the color space dividing unit 405 correspond to, that is, in which class on the color frequency distribution. Determine whether to add. In S4085, the class of the color frequency distribution corresponding to the luminance level and hue level determined in S4084 is called, and 1 is added to the number of pixels stored in this class.

  In S4086, the color information of P_ds is added to the color information of the corresponding class. In S4087, the position information of the corresponding class is updated based on the pixel position of P_ds. If the coordinate information of the pixel position is retained as the position information of the class as it is, the memory size becomes large. Therefore, in this embodiment, the position information is represented by a 16-bit area code. This is stored in the location information in the histogram visualized in FIG.

  Details of the area code will be described with reference to FIG. FIG. 9A illustrates the area code setting range on the tile. N represents the number of pixels on one side of the smoothing tile, and each tile is divided as shown in the figure (areas 901 to 916). In the present embodiment, assuming that the size of the processing target tile of the histogram is N, the tile boundary portion is set to N / 8 from the tile end, and 16 regions obtained by dividing the tile central portion into four equal portions are set. That is, here, the tile boundary portions are the areas 901 to 905, 908, 909, and 912 to 916, and the tile central portion is the areas 906 to 911. For example, when a smoothing tile having a size of 16 × 16 pixels is targeted, an area obtained by dividing the vertical direction and the horizontal direction at a pixel ratio of 2: 6: 6: 2 is set.

  Note that the size of each area is different in order to consider the handling of a very small area in which a part of a character or the like is divided by tile division and exists very small in the tile. Therefore, at least the tile boundary and the tile center need only be defined, and the range (size) of the area to be divided is not limited to this. For example, a portion in contact with a tile boundary line among 16 equally divided tiles may be set as a tile boundary portion, and the other portion may be set as a tile central portion. Similarly, in FIG. 9A, the tile boundary part and the tile center part are composed of a plurality of areas, but the number of areas to be divided is not limited to this, for example, the tile boundary part and the tile center part are Each may be defined as one area (two areas in total).

  As shown in FIG. 9B, the location information corresponding to each area defined in FIG. 9A is defined by a 16-bit area code. Here, every 4 bits out of 16 bits are defined as MIDDLE AREA, TOP / BOTTOM ENDS, RIGHT / LEFT ENDS, and CORNER. MIDDLE AREA corresponds to areas 906, 907, 910, and 911 in FIG. 9A, respectively. Similarly, TOP / BOTTOM ENDS corresponds to areas 902, 903, 914, and 915, RIGHT / LEFT ENDS corresponds to 905, 908, 909, and 912, and CORNER corresponds to 901, 904, 913, and 916, respectively.

  In S4087, the pixel position of P_ds is referred to, it is determined which of the areas 901 to 916 shown in FIG. 9A enters, and the bit of the corresponding area code is updated from 0 to 1. For example, when the pixel position of P_ds recorded in a certain class corresponds to the area 901 in the area code setting range, the fourth bit from the right corresponding to the area 901 is updated from 0 to 1 as shown in FIG. . When the processing from S4083 to S4087 is repeated for the pixels of the smoothed tile (S4082), the position information of a certain class is as shown in FIG. 9D, for example. This means that the pixels included in this class existed in the areas 909, 910, 913, and 914 in the area code setting range.

  S4088 means that the process of S4089 is repeated for the number of classes. In S4089, the average of the color information of the pixels counted in the color frequency distribution class is obtained. Specifically, the sum of the color information added to each class in S4086 is divided by the number of pixels of each class. Implement this process for all classes. Through the above processing, the average color information for each class of the color frequency distribution (histogram) and the position information (area code) for each class of the color frequency distribution are generated from the smoothed tile.

  Next, details of the representative color selection unit 409 will be described. FIG. 10 is a flowchart of the representative color selection unit 409. In S4091, a method for determining a representative color is selected. Here, the representative color determination method is selected from the distribution shape of the histogram. For example, when all of the pixels in the tile are included in a certain class in the histogram, the process shifts to a process for making the tile one color (method 1, first representative color processing means). Further, if the histogram includes a predetermined number or more (for example, 90% or more) of pixels in a tile in a certain class and the remaining pixels are also recorded in a nearby class, the representative color of the tile is 1 Since there is a high possibility of being a color, the process shifts to a process for making a tile one color (method 1).

  Pixels that exist only at the same hue level, but histograms with multiple local maxima in the luminance level direction and histograms with pixels in different hue level classes may have two or more colors in the tile. High nature. Therefore, the process shifts to processing for selecting two or more colors as representative colors from the tile (method2, second representative color processing means).

  Further, in the case of a histogram having the same hue level but a narrow distribution width in the luminance level direction (that is, the luminances of the pixels in the block are all approximated), the color unevenness of the halftone dot region is left or left. There is a possibility that the color information of the minimal region should be mixed. Therefore, the process shifts to a process of selecting one or two representative colors from the tile (method 3, third representative color processing means). Here, the method branches to three methods, but a more suitable color reduction method may be prepared and selected by looking at the histogram and other feature amounts.

  In S4092, the representative color candidate class is determined. Here, the branching is performed corresponding to the representative color determination method selected in S4091. Since method 1 is a process for determining one representative color, the class having the largest number of pixels is set as the representative color candidate class. Method 2 is a process for determining two or more representative colors. Here, the class having the maximum color frequency is set as the representative color candidate class. Note that the representative color candidate class determination method described here is not limited to this, and other methods may be used as long as they are image processing methods using a general histogram or other feature amounts. In method 3, the representative color is determined to be one color or two colors. In order to determine whether a tile to be processed has only a halftone dot area or a minimal area cut off at the end of the tile, information on peripheral tiles is generally used. However, there is a high possibility that the color information of adjacent tiles other than the corresponding tile and the background information of the entire page cannot be referred due to the memory size. Therefore, method 3 determines the representative color to be one color or two colors only with the histogram information of the processing target tile.

[Representative color determination method (method 3)]
Details of the method 3 for determining the representative color of method 3 described above will be described with reference to FIG. In S40921, the class having the largest number of pixels in the histogram is set as “candidate 1” as the representative color. In S40922, among the classes in which the pixels are counted, the class farthest from the candidate 1 class in the luminance level direction is set as the “candidate 2” representative color. The background color and the foreground color such as characters use a feature that tends to make a difference in brightness so that it can be easily seen by human eyes. In S40923, it is determined whether the number of pixels of the candidate 2 class is sufficiently large. Here, a threshold value that is a criterion for the number of pixels is defined in advance, may be determined empirically, or may be derived from an arbitrary calculation formula. If it is determined that the number of pixels is sufficiently large, the process proceeds to S40925; otherwise, the process proceeds to S40924. In S40924, it is determined whether or not candidate 2 is color information to be left (necessary minimum region color determination). Specifically, it is determined whether or not the area code is only a tile boundary by referring to the position information of the candidate 2 class.

  Here, the tile boundary portion means areas 901 to 905, 908, 909, and 912 to 916 in FIG. 9A. There is a possibility that a small character area partially broken from an adjacent tile by tile division exists only at the tile boundary. That is, if the area code of the candidate 2 class is only the tile boundary, this class can be regarded as the color of the minimal area to be left. Therefore, if it is determined Yes in S40924, the process proceeds to S40925, and the candidate 2 class is adopted. On the other hand, when the number of pixels is small and the area code corresponds to any of the areas 906, 907, 910, and 911 in the center of the tile, the color of the candidate 2 class is a part of the halftone dot composition point, It can be regarded as a color (noise) that should not be left behind. Therefore, if it is determined No in S40924, the process proceeds to S40926, where the candidate 2 class is rejected from the representative color. As described above, the representative color candidate classes are determined by methods 1 to 3, and the process proceeds to the next process (FIG. 10, S4093).

  In S4093, the color information of the representative color is determined. Details will be described with reference to FIG. S40931 means that the processing of S40932 to S40934 is repeated for the representative color candidate classes. In S40932, it is determined whether the color information of the candidate class is used as it is or is integrated with the color information of the adjacent class (class color information integration determination). Specifically, the candidate class and the non-candidate class adjacent to this class are referred to, and it is determined whether the position information of these classes matches. In some cases, the color information of halftone dots is not distributed in a single class but is distributed in adjacent classes. Then, the candidate class determined in S4092 (the most frequent class in the histogram, or the class that becomes the maximum point) is shifted for each tile, and if the average color information of the class is used as the representative color of the tile as it is, the color difference between the tiles May become large.

  In general, correction is performed to reduce the color difference between the tiles by referring to the color information of the background of the surrounding tiles and the entire page, but in this embodiment, the problem is solved only by the histogram information of the corresponding tiles. That is, the class having the color information of the constituent points of the halftone dots existing in the same halftone dot area is searched from the histogram, and is integrated with the color information of the candidate class. In order to determine whether or not the class has color information of the constituent points of the halftone dots, whether or not pixels in the same area as the candidate class are added may be used as a determination criterion. That is, it can be determined whether the class area codes are the same. Therefore, when the area codes match in S40932 (in the case of Yes), the process proceeds to S40933, where a color obtained by integrating the color information of each class is recalculated, and this color information is adopted as the representative color. The calculation method of the integrated color may be a simple average of RGB values in each class, or a weighted average with the number of pixels added to the class as a weight. On the other hand, if it is determined No in S40932, that is, if the area codes of the adjacent classes do not match, the process proceeds to S40934, and the color information of the candidate class itself is adopted as the representative color. The representative color information 410 (FIG. 4) is determined by the above processing, and the process proceeds to the color replacement unit 411 (FIG. 4) of the next processing.

  As described above, in the present embodiment, the position information of each class of the histogram is given when generating the histogram, and the position information of the histogram is used when selecting the representative color. Thereby, it is possible to accurately determine whether the representative color candidate class is one color or two colors only with the histogram information of the processing target tile. Similarly, the representative color of the halftone dot region can be determined more accurately only with the histogram information of the processing target tile, and the color difference between the tiles can be reduced.

<Second Embodiment>
In the first embodiment, the position information of the histogram is used in method 3 (one or two representative colors are selected from tiles in a halftone dot area or tiles including a minimal area). However, the position information of the histogram is also effective in method 2 that determines the representative color of the tile to be two or more. For example, after calculating a candidate class from the histogram, position information is used to determine whether the candidate class is a color to be left. If the area code is only the tile boundary, it is adopted for the representative color candidate class, otherwise it is rejected.

  As described above, even in a tile having two or more representative colors, it is possible to accurately determine whether or not a candidate color should be left based only on the histogram information of the processing target tile.

<Other embodiments>
The present invention can also be realized by executing the following processing. That is, software (program) that realizes the functions of the above-described embodiments is supplied to a system or apparatus via a network or various storage media, and a computer (or CPU, MPU, or the like) of the system or apparatus reads the program. It is a process to be executed.

Claims (10)

Tile dividing means for dividing the input image data into tiles of a prescribed size;
Color frequency distribution generating means for generating a color frequency distribution of pixels included in the divided tile for each of the divided tiles;
Position information giving means for giving position information of pixels in the tile belonging to each class to each class of the generated color frequency distribution;
Representative color selection means for selecting a representative color in the tile;
Color replacement means for replacing color information of pixels of the tile with the representative color selected by the representative color selection means;
The representative color selection means is:
Necessary minimum area color determination means for determining a representative color candidate from color information of the pixel using the position information in the color frequency distribution class;
An image processing apparatus comprising: class color information integration means for integrating the color information of the class that is the representative color candidate using the position information in the class of the color frequency distribution and determining the representative color.   The image processing apparatus according to claim 1, wherein the color frequency distribution is information calculated based on a relationship between a hue and a luminance of a pixel included in the tile.   3. The image according to claim 1, wherein the position information provided by the position information adding unit is defined as an area code indicating a position of an area obtained by dividing the tile by a specified number. 4. Processing equipment. The representative color selection means is a means for determining the representative color candidates.
First representative color processing means for determining the representative color candidate as one color;
Second representative color processing means for determining the representative color candidates as a plurality of colors;
And third representative color processing means for determining the representative color candidate as one or two colors,
In the color frequency distribution, the first representative color processing means is configured such that all pixels included in the tile are located in the same class, or a predetermined number or more of pixels are located in the same class and are close to the class. Selected when the remaining pixels are in the class
In the second representative color processing means, in the color frequency distribution, all the pixels included in the tile have the same hue and a plurality of pixels having the same luminance, or the pixels included in the tile are different. Selected if it has a hue,
The third representative color processing means is selected when all the pixels included in the tile in the color frequency distribution have the same hue and the respective luminances are approximate. Item 4. The image processing device according to any one of Items 1 to 3. Smoothing means for generating a smoothed tile by smoothing the tile divided by the tile dividing means;
The tile used in the color frequency distribution generation means, the position information giving means, and the representative color selection means is a smoothing tile generated by the smoothing means,
The image processing apparatus according to claim 1, wherein the tile used in the color replacement unit is a tile before the smoothing process. A tile dividing step in which the tile dividing means of the image processing apparatus divides the input image data into tiles of a prescribed size;
A color frequency distribution generating step of generating a color frequency distribution of pixels included in the divided tile for each of the divided tiles;
A position information providing step in which the position information providing unit of the image processing apparatus assigns the position information of the pixels in the tile belonging to each class to each class of the generated color frequency distribution;
A representative color selection unit of the image processing apparatus selects a representative color in the tile;
A color replacement step of replacing the color information of the pixels of the tile by the representative color selected in the representative color selection step;
The representative color selection process includes
A necessary minimum area color determining unit of the image processing apparatus, which uses the position information in the color frequency distribution class to determine a representative color candidate from color information of the pixel;
A class color information integration step in which the class color information integration unit of the image processing apparatus integrates the color information of the class that is the representative color candidate using the position information in the class of the color frequency distribution and determines the representative color; A control method characterized by comprising: Computer
Tile dividing means for dividing the input image data into tiles of a prescribed size;
Color frequency distribution generating means for generating a color frequency distribution of pixels included in the divided tile for each of the divided tiles;
Position information giving means for giving position information of pixels in the tile belonging to each class to each class of the generated color frequency distribution;
Representative color selection means for selecting a representative color in the tile;
The representative color selected by the representative color selection means functions as a color replacement means for replacing the color information of the pixel of the tile,
The representative color selection means is:
Necessary minimum area color determination means for determining a representative color candidate from color information of the pixel using the position information in the color frequency distribution class;
A program that functions as a class color information integration unit that integrates color information of a class that is a representative color candidate using the position information in the class of the color frequency distribution and determines the representative color. Tile dividing means for dividing the input image data into tiles of a prescribed size;
Color frequency distribution generating means for generating a color frequency distribution of pixels included in the divided tile for each of the divided tiles;
Position information giving means for giving position information of pixels in the tile belonging to each class to each class of the generated color frequency distribution;
Representative color selecting means for selecting a representative color in the tile based on the generated color frequency distribution and the given position information;
An image processing apparatus comprising: a color replacement unit that replaces color information of pixels of the tile with the representative color selected by the representative color selection unit.   The representative color selection means selects a representative color in the tile based on the color frequency distribution and whether the position information given to the class that is a representative color candidate is a boundary portion of the tile. The image processing apparatus according to claim 8.   In the case where the representative color selecting means includes the same position information as the position information given to a class that is a representative color candidate and the position information given to a class adjacent to the class that is the representative color candidate. The image processing apparatus according to claim 8, wherein the representative color is determined by integrating the color information of the class that is the representative color candidate and the class adjacent thereto.

Images