JP2013102300A - Image processor and image processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily and accurately select a character color from a histogram in which blurring becomes dominant in color reduction processing.
SOLUTION: A plurality of classes configured based on a plurality of areas set in a hue circle and a plurality of luminance levels set in a luminance direction are determined, and a frequency distribution of each class is created from an input image, and It is determined whether the hue difference is equal to or greater than a predetermined threshold. Here, when it is determined that the threshold value is equal to or greater than the predetermined threshold, the color information of the class having the maximum hue and the color information of the class having the minimum hue are selected as the representative colors. On the other hand, when it is determined that it is not equal to or greater than the predetermined threshold, the color information of the class having the highest frequency distribution and the color information of the class having the most luminance difference from the maximum class are selected as representative colors. Then, each pixel value included in the input image is replaced with the selected representative color.
[Selection] Figure 2

Description

  The present invention relates to an image processing apparatus and an image processing method for performing a color reduction process by selecting a representative color of an input image.

  In recent years, the digitization of information has progressed, and instead of storing a paper document, a system that reads a paper document with a scanner or the like, stores it as electronic data, and transmits the electronic data to another device has become widespread. In addition, the color used for the paper document itself has been changed from black and white to color (multi-value).

  An electronic document stored as electronic data is required to have a clear color for the character portion and a uniform color for the background 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 halftones using a screen or the like, blurring and unevenness due to the colors of the screen components (called halftone dots) themselves occur, which affects readability and image quality.

  For this reason, in document digitization processing, color reduction processing is generally performed in which a representative color is selected from an image and reduced.

  It is assumed that the color reduction processing of the scanned image is generally incorporated in a device such as a scanner or a multifunction machine and is operated in real time. Therefore, the color reduction processing requires high speed and memory saving as well as high accuracy.

  A technique for determining a blur color generated at an area boundary in a color reduction process in units of tiles is disclosed (for example, see Patent Document 1). First, an input image is divided into tiles of a prescribed size, and a low-resolution image for generating a color frequency distribution and an edge image for generating an edge frequency distribution are generated. When a pixel in a low-resolution image is added to a class (referred to as bin) having a first frequency distribution (color frequency distribution), if the pixel in the image at the same position as the added pixel is an edge region, the second Add to the frequency distribution (edge frequency distribution).

  Of the two frequency distributions created in this way, the top three colors of the first frequency distribution are set as candidate colors, and if the second frequency distribution of the candidate colors is large, that color is the color at the boundary of the region. In other words, it is judged as blurring and excluded from the candidate colors.

JP 2010-268438 A

  In general, when the boundary length between the background and the foreground in the tile is short, the appearance frequency of the boundary blur generated by mixing both colors is smaller than the foreground and background colors. Therefore, since a color with a high appearance frequency in the tile image can be regarded as an important color that becomes a foreground or background in the tile image, it is common to select a color with a high frequency as a representative color.

  On the other hand, as shown in (F) of FIG. 5, in tiles with many foreground and background boundaries that tend to appear as part of complex Chinese characters, the border blurs more than the color representing the foreground and background itself. May appear more frequently.

  Therefore, if the selection criterion of the representative color is only the appearance frequency of the pixel, a color that should not be selected, that is, blur may be the representative color.

  Further, in the technique of Patent Document 1, a separate memory for using the edge image and the second frequency distribution is required to determine color bleeding.

  The present invention provides an apparatus and method that can easily and accurately select a character color from a histogram in which blurring has become dominant.

The image processing apparatus of the present invention sets a plurality of regions in the hue circle by dividing a hue circle constituted by saturation and hue based on a predetermined criterion, and further, brightness of a predetermined color space Color space dividing means for setting a plurality of luminance levels by dividing a direction and defining a plurality of classes configured based on a plurality of areas set in the hue circle and a plurality of luminance levels set in the luminance direction When,
For all the pixels included in the first image data, a frequency distribution is created by counting the number of pixels belonging to each class determined by the color space dividing means, and the colors of the pixels belonging to each class A frequency distribution creating means for obtaining an average value of information as color information of each class;
Determination means for determining whether or not a hue difference between classes is greater than or equal to a predetermined threshold based on the color information of each class obtained by the frequency distribution creating means;
If the determination means determines that the hue difference between the classes is greater than or equal to the predetermined threshold, the color information of the class having the maximum hue and the color information of the class having the minimum hue are selected as representative colors. When the determination means determines that the hue difference between the classes is not equal to or greater than the predetermined threshold, the frequency distribution created by the frequency distribution creation means is the largest luminance difference from the color information of the maximum class and the maximum class. Representative color selection means for selecting color information of a certain class as a representative color;
Replacement means for replacing each pixel value of a pixel included in the first image data using the selected representative color;
It is characterized by having.

  According to the present invention, it is possible to easily and accurately select a character color from a histogram in which bleeding is dominant in the color reduction processing.

1 is a block diagram showing the overall configuration of an image processing system according to a first embodiment. 2 is a block diagram showing a functional configuration of the image processing apparatus 100. FIG. (A) is a figure which shows an example of an input image, (B) is a figure which shows a some tile image. The figure for demonstrating HSV color space. (A)-(C) are figures which show a hue circle, (D)-(F) is a figure which shows a histogram. (A) is a figure which shows frequency distribution, (B) is a figure which shows the data structure of frequency distribution. The flowchart which shows a process in case bin exists in several hue ranges. The flowchart which shows a process in case bin exists only in the same hue range. Flowchart showing representative color selection processing according to the second embodiment

  Hereinafter, embodiments for carrying out the invention will be described in detail with reference to the drawings. The following embodiments are merely examples, and are not intended to limit the scope of the present invention.

[First Embodiment]
First, as a first embodiment, an image processing system in which an image processing apparatus according to the present invention and a personal computer (PC) are connected to each other via a network will be described as an example.

<System configuration>
FIG. 1 is a block diagram showing the overall configuration of the image processing system according to the first embodiment. The image processing apparatus 100 includes a scanner 101 that reads a document image and generates image data, and a network I / F 105 for inputting / outputting data to / from an external apparatus. On the other hand, the 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.

  The image processing apparatus 100 includes a CPU 102 that executes a processing program for performing image processing on image data read by the scanner 101 or image data acquired from an external apparatus via the network I / F 105. In addition, the image processing apparatus 100 includes a work memory when executing the processing program, a memory 103 used for temporary storage of data, and a hard disk 104 that stores the program and data.

  FIG. 2 is a block diagram illustrating a functional configuration of the image processing apparatus 100. The processing unit described below is assumed to be realized by the CPU 102 executing various processing programs. However, a part or all of the processing unit may be configured by an electric circuit (ASIC or the like). .

  In FIG. 2, an image input unit 201 inputs image data read by the scanner 101 or image data acquired from an external device via the network I / F 105. Here, it is assumed that the input image data is a color image that is expressed in an RGB color space and is generated with a resolution of 600 dpi. The pixel value is expressed by 24 bits (8 bits for each color), and can take values of 0 to 255 for each color.

  FIG. 3A is a diagram illustrating an example of an input image input from the image input unit 201. The color information of the input image 300 is different for the background area 301, the character area 302, and the image area 304, and the area 303 is a blur that occurs at the boundary between the background area 301 and the character area 302. An area 305 is a blur that occurs at the boundary between the background area 301 and the image area 304. Here, the background area 301 is chromatic red, the character area 302 is chromatic red that is darker than the background area 301, the image area 304 is chromatic yellow, and the color is represented by halftone dots. Note that the blur originates from what originally existed in the document (generated during printing) or occurs when the scanner 101 reads the document image.

  Returning to FIG. 2, the tile dividing unit 202 divides the image data input from the input unit 201 into a plurality of tile images (first image data) having a predetermined size. By dividing into tiles and processing sequentially, the amount of memory required for processing can be reduced. (B) shown in FIG. 3 is a diagram showing a plurality of divided tile images. In this example, the input image 300 is divided into 18 parts with 32 pixels square as one unit. The following processing is executed for the input tile 203 divided into tiles.

  For the following explanation, numbers are assigned to the vertical and horizontal tiles, and the tile positions are defined by (vertical number, horizontal number). In the example of (B) shown in FIG. 3, the position of the leftmost and uppermost tile is (0, 0), and the position of the rightmost and lowermost tile is (2, 5).

  Here, referring back to FIG. 2, the histogram generation unit 204 includes a color space division unit 205 and a frequency distribution generation unit 206. Since the color space used in the color space dividing unit 205 is an HSV color space, the color space dividing unit 205 includes conversion processing from the RGB color space to the HSV color space. Details of the histogram generation unit 204 will be described later.

  The one-color / multicolor determination unit 207 determines whether the tile image has only the background (one color) or has the foreground (multicolor). Here, reference is made to the histogram variance (number of bins with added pixels, number of hue ranges in which bins exist and number of luminance levels), and if the variance is greater than a predetermined value, it is multicolor, otherwise 1 Color. Here, bin is a class in the histogram expressing the frequency distribution.

  The representative color selection unit 208 determines color information (representative color information 210) of the representative color in the tile from the histogram generated by the histogram generation unit 204. The representative color selection unit 208 includes a maximum frequency bin determination unit 209. Details of the representative color selection unit 208 will be described later.

  The color replacement unit 211 obtains the distance between the color information of the pixel unit of the input tile 203 and the representative color information 210 determined by the representative color selection unit 208, and assigns the color to the representative color closest to the input image. Perform replacement quantization. As a result, the quantized tile 212 that has been subjected to the color reduction processing for each tile is generated.

  In this example, the color replacement unit 211 calculates the Euclidean distance between the target pixel and the representative color RGB space, but the color space and the distance calculation method are not limited to this.

  Next, the histogram generation unit 204 and the representative color selection unit 208 of the image processing apparatus according to the present invention will be described in detail. First, processing by the color space dividing unit 205 of the histogram generating unit 204 will be described.

<Color space division>
The color space dividing unit 205 determines a plurality of bins used for generating the frequency distribution information of the frequency distribution creating unit 206.

  First, the hue (H) and saturation (S) values in the HSV color space and the luminance (Y) value in the YUV color space are derived from the RGB pixel values corresponding to each pixel. The HSV color space is shown as a cone composed of hue (Hue), saturation (Saturation), and lightness (Value) as shown in FIG. Here, the radial direction is saturation and is represented by a value of 0 to 1 corresponding to the distance from the center line of the cone. The center of the circle (S = 0) is an achromatic color, and the far side (value close to 1) is a more vivid chromatic color. The hue corresponds to the circumferential direction (the angular direction of the circle), and the hue is expressed by a value of 0 to 360 degrees. For example, 0 degree corresponds to a red color, 50 degrees corresponds to a yellow color, 120 degrees corresponds to a green color, and 200 degrees corresponds to a blue color. The lightness V indicates the brightness of the color and is indicated by a value from 0 to 100 (%), and becomes brighter as the value increases. Note that 0% corresponds to black and 100% corresponds to white.

  When the maximum value of the RGB pixel values (R, G, B) is MAX and the minimum value is MIN, the HSV pixel values (H, S, V) are calculated as follows.

H = 60 × (GB) / (MAX−MIN) +0 if MAX = R
H = 60 × (BR) / (MAX−MIN) +120 if MAX = G
H = 60 × (RG) / (MAX−MIN) +240 if MAX = B
S = (MAX-MIN)
V = MAX
Here, each value of RGB shall be represented in the range of 0-255. It should be noted that the luminance Y in the YUV color space is closer to the sense of brightness when viewed by human eyes than the brightness V in the HSV color space. For this reason, in the first embodiment, the luminance Y of the YUV color space is used instead of the brightness V.

  That is, in the first embodiment, the hue and saturation of the HSV color space and the luminance of the YUV color space are used. Hereinafter, the color space represented by the hue and saturation of the HSV color space and the luminance of the YUV color space is referred to as a Y / HS color space. The value of luminance Y is calculated from the RGB pixel values (R, G, B) as follows.

Y = 0.29891 × R + 0.58661 × G + 10.1448 × B
With the above formula, the values of H, S, and Y can be calculated from the RGB pixel values.

  Next, the color space is divided by setting a threshold value for hue and saturation and a threshold value for luminance. Since the luminance direction is one-dimensional linear, the luminance value may be divided based on a predetermined luminance threshold. Hereinafter, a case where the luminance level is set to 16 levels will be described as an example. At this time, the area is divided into 16 (P + 1) areas (Y0 to Y15) based on 15 (predetermined P) brightness thresholds between the brightness values.

  The lower part of FIG. 4 is a diagram illustrating an example in which the hue ring 401 expressed by hue and saturation is divided into a plurality of regions 402 to 406. First, the hue circle 401 is divided into an achromatic region (low saturation region) and a chromatic region (high saturation region). As described above, the saturation can be expressed by the radial distance in the hue circle.The closer to the origin of the hue circle, the closer to the achromatic color such as white, gray, black, etc. Become. Therefore, a predetermined saturation threshold S ′ (predetermined threshold) is set, and a range where the saturation is smaller than the saturation threshold S ′ is defined as an achromatic region 402 (H0).

  Next, division of the chromatic color area will be described. Since the hue is expressed by the phase in the hue circle, the hue circle is divided at a set angle to divide the chromatic color region. In the first embodiment, the color space is divided into four equal parts, that is, the divided hue values are set at intervals of 90 degrees, and the chromatic color regions 403 to 406 are set to H1 to H4.

  The H1 divisional hue value used as a reference for hue division is obtained by sequentially scanning pixels included in the tile image and adding ± 45 degrees to the hue value of a pixel having a saturation equal to or higher than the saturation threshold S ′ that is first discovered. Shall be. Hereinafter, a specific example will be described.

  (A) shown in FIG. 5 is a hue circle of the tile image (0, 0) shown in FIG. First, the H1 segmented hue value sequentially scans the tile image (0, 0), and when a pixel having a saturation threshold value S ′ or more is found, the pixel value 501 of the first found pixel (reference pixel). The division position of the hue circle is determined at ± 45 degrees with respect to the center. Then, the divided hue values are set at intervals of 90 degrees to divide into chromatic color regions H1 to H4. A color distribution 502 indicates a distribution of pixel values after all tile images (0, 0) are scanned. The tile includes a light red background region 301 and a dark red character region 302, so there is no significant difference in hue between the two regions. Therefore, the color distribution 502 on the hue circle includes the pixel values of both the background area 301 and the character area 302.

  (B) shown in FIG. 5 is a hue circle of the tile image (0, 3) shown in FIG. First, the segmented hue value of H1 is determined by scanning the tile image (0, 3) in order, and is determined at ± 45 degrees centering on the pixel value 503 equal to or greater than the first discovered saturation threshold S ′. Divided. Color distributions 504, 505, and 506 indicate pixel value distributions after the tile image (0, 3) has been scanned. Since the tile includes a light red background region 301 and a yellow image region 304, the color distribution 504 on the hue circle is the background region 301, and the color distribution 505 is the image region 304. Since there is a difference in hue between red in the background and yellow in the foreground, the color distribution 504 and the color distribution 505 are separated from each other in the hue circle. However, since there is a blurred area 305 generated at the boundary between the background and the character, a blurred color distribution 506 exists on the hue ring so as to connect the color distribution between the character and the background.

  (C) shown in FIG. 5 is a hue circle of the tile image (1, 4) shown in FIG. First, the segmented hue value of H1 is determined at ± 45 degrees centering around the pixel value 507 equal to or higher than the saturation threshold S ′ that is found first by sequentially scanning the tile images (1, 4). Divided. Color distributions 508, 509, and 510 indicate pixel value distributions after the tile images (1, 4) have been scanned. The tile image (1, 4) includes a yellow image region 304 in a light red background region 301, like the tile image (0, 3) described in FIG. 5B. The color distribution 508 on the hue circle corresponds to the background area 301, the color distribution 509 corresponds to the image area 304, and the color distribution 510 corresponds to the blur area 305.

  In the following description, each region (achromatic color region H0 and divided chromatic color regions H1 to H4) when the hue ring 401 expressed by hue and saturation is divided is grouped into hues. This is called a range.

<Generation of frequency distribution information>
First, the frequency distribution generation unit 206 divides the luminance level Y (Y0 to Y15) divided in the luminance direction by the color space division unit 205 and the hue ring represented by hue and saturation, and the hue range H (H0). To H4). The number of appearances of pixels in the tile image included in the array element (= class (bin)) of each two-dimensional array is counted (counted).

  FIG. 6A shows an arrangement of frequency distributions. In each bin of the frequency distribution of (A), the number of appearances of a tile image having luminance information and hue range color information corresponding to each bin, and color information (hue, saturation) corresponding to the counted pixels are included. , RGB) average values.

  (B) shown in FIG. 6 is a diagram showing a data structure of a frequency distribution. The frequency distribution holds the hue value of the segment of the chromatic color area defined by the bin arrangement and the color space division. In creating the frequency distribution, it is determined to which bin the color information of the pixel applies, and the pixel appearance count of the corresponding bin is added (incremented). At the same time, the pixel color information is added to the bin color information.

  After determining the pixels of all tile images, the color information of each bin is divided by the number of pixel appearances. That is, the average value of the color information of the pixel added to bin is calculated, and this value is determined as the color information of the bin.

  (D) to (F) shown in FIG. 5 show examples of the frequency distribution of each bin defined by the hue range and the luminance level. The frequency distribution created in the first embodiment is equivalent to creating a luminance histogram of five hue ranges per tile image. In addition, since the color of the input image is represented by halftone dots, even if only one color is originally visible to the human eye, the color information for each pixel varies due to the influence of the halftone dot composing points. For this reason, in a histogram, pixels are generally added to bins having a plurality of luminance levels.

<Determination of representative colors>
The representative color selection unit 208 determines a representative color based on the information generated by the frequency distribution generation unit 206. If the one-color / multi-color determining unit 207 determines that one color, that is, no foreground exists, the representative color of the tile image may be the average value of the color information of the pixels of the tile image. Here, the details of the representative color determination in the case where there are multiple colors, that is, the foreground, in the one-color / multi-color determination will be described.

  In general, a color having a high appearance frequency in a tile image can be regarded as a background representative color of the tile image. In addition, printed materials are often designed with a difference in brightness and hue between the foreground and background so that they can be easily seen by the human eye. Can be adopted.

From the above, the characteristics of bins to be used for calculating representative colors are summarized below.
* Frequent bin
* Bin that has a difference in luminance level compared to bins with high frequency
* Bin with different hue range
In the first embodiment, two types of algorithms are prepared in order to switch the representative color selection algorithm depending on the number of hue ranges in which pixels are added on the histogram.
(1) When bins exist in a plurality of hue ranges (2) When bin exists only in the same hue range Details of the above two types of algorithms will be described in order. First, the algorithm (1) will be described in detail. A tile having a histogram in which bins exist in a plurality of hue ranges is a state in which the hue of the background color and the foreground color of the input image are different and each color is distributed to different hue ranges when the color space is divided ( Hue ring: (B) of FIG. When viewed from the histogram, as shown in FIG. 5E, the background color bin and the foreground color bin appear in different hue ranges (H1 and H2 in FIG. 5E).

  Therefore, in the case of a histogram having bins in a plurality of hue ranges, a bin having the maximum frequency in each hue range and a bin having a difference in luminance level from the bins may be selected as representative colors.

  A representative color selection process when bins exist in the plurality of hue ranges described above will be described with reference to FIG. This process is performed for each hue range in which bins exist. First, in S701, the bin color information having the maximum frequency in a certain hue range is adopted as the representative color, and the process proceeds to S702. In S702, it is determined whether or not there is a difference in the luminance levels of the colors included in the hue range. Specifically, in the achromatic color region H0, the determination may be made based on the number of bins obtained by adding pixels in the range of H0.

  Further, the criterion for the presence or absence of a luminance difference in the chromatic color regions H1 to H4 is different from that of the achromatic color region. This is because halftone dots are used to represent chromatic colors in a printed matter, so that even if one color is desired to be represented, a difference in luminance is likely to occur in color information on a pixel basis due to the composition of the halftone dots. For this reason, it is impossible to determine whether or not there are a plurality of bins that should be representative colors in the hue range only by the number of bins to which pixels are added. Therefore, in order to confirm whether or not there is a difference in luminance, whether or not bins exist continuously is used as a criterion. That is, if bins are continuous in the hue range, it is determined that there is no difference in luminance level. Conversely, if bins are present in the stepping stone, this hue range is regarded as having a difference in luminance level.

  If it is determined in S702 that there is a difference in luminance level (YES), the process proceeds to S703, and if it is N0, the representative color selection in this hue range is completed, and the process proceeds to representative color selection in another hue range. In S703, bin color information having the greatest difference in luminance level within the same hue range with respect to the bin with the highest frequency selected in S701 is adopted as the representative color. Specifically, it is the bin that is located farthest from the maximum frequency bin in the luminance direction.

  When the above-described representative color selection processing is applied to the histogram (E) shown in FIG. 5 of the tile image (0, 3), it is determined that there is no difference in luminance level between H1 and H2, so the representative color to be selected is H [1] Y [7] and H [2] Y [11].

  Next, the algorithm (2) will be described in detail. A tile having a histogram in which bins exist in the same hue range is a hue that is close or achromatic, and a combination of foreground and background colors with different densities (hue ring: FIG. 5A) Is assumed. Looking at the histogram, as shown in FIG. 5D, the foreground color bin and the background color bin appear in the same hue range. When bins appear only in the same hue range, the maximum frequency bin in the hue range is the background color, and the bin having the most luminance difference from the background color is the foreground character color.

  From the above, if the bin exists only in the achromatic hue range H0, the bin having the highest frequency level and the bin having the most difference in luminance level may be selected as the representative color.

  However, when the hue circle is generated in the boundary between the foreground and the background as shown in FIG. 5C and the histogram in FIG. The foreground color and the background color that should be divided into different hue ranges may be arranged in the same hue division.

  In such a case, there are cases where the same processing as in the case where the hues of the foreground and the background shown in FIG. Because the shape of the foreground area in the tile is complicated, the boundary area between the background and the foreground is increased, and the frequency of the blurring color of the boundary is higher than the frequency of the color representing the foreground and the background itself. It is a case.

  In order to deal with this case, when a representative color is selected from a histogram in which bins appear only in the same chromatic hue range, it is determined whether or not the bin with the highest frequency should be the representative color. Details of the representative color determination process when bin appears only in the same chromatic hue range H1 will be described with reference to FIG. S801 refers to the color information of each bin and determines whether there is a difference in hue. Specifically, the bin of the maximum hue value and the bin of the minimum hue value are extracted, and if the difference is equal to or larger than a predetermined value (greater than a predetermined threshold), it is determined that there is a hue difference between classes.

  In S801, if the hue difference between classes is greater than or equal to a certain level (if YES), the process proceeds to S802, and the bin with the largest hue and the bin with the smallest hue are selected as representative colors. However, if NO, the process proceeds to S803, and the bin having the highest luminance difference is selected as the representative color from the maximum frequency bin and the maximum frequency bin.

  The histograms (D) and (F) shown in FIG. First, in the case of the histogram shown in FIG. 5D, the difference in hue is small as shown in FIG. 5A. Therefore, the process proceeds to S803, where H [1] Y [7] having the maximum frequency is selected as the representative color, and H [1] Y [3] having a difference in luminance level is selected as the representative color.

  Next, in the case of the histogram of (F) shown in FIG. 5, the difference in hue is large as shown in (C) of FIG. For this reason, the process proceeds to S802, and representative colors are selected from H [1] [6] and H [1] [12] having the maximum and minimum hues as representative colors.

  According to the first embodiment, it is possible to select a representative color with high accuracy by simple processing even for an image in which blurring is dominant.

[Second Embodiment]
Next, a second embodiment according to the present invention will be described in detail with reference to the drawings. In the first embodiment, when bins exist only in the same chromatic hue range H1, it is determined whether or not the maximum frequency bin should be adopted as the representative color based on the hue difference of bins. In the second embodiment, even when there are bins in a plurality of hue ranges, it is determined whether or not the maximum frequency bin should be adopted as the representative color.

  FIG. 9 is a flowchart showing a representative color selection process when bins exist in a plurality of hue ranges. This process is performed for each hue range in which bins exist. First, in step S <b> 901, it is determined whether there is a difference in hue by referring to color information of bins included in a certain hue range. Specifically, the maximum hue value bin and the minimum hue value bin included in the same hue range are extracted, and if the difference is greater than or equal to a certain level, it is determined that there is a hue difference between classes.

  If the hue difference between the classes is large in S901 (if YES), the process proceeds to S902, where the bin with the largest hue and the bin with the smallest hue are selected as representative colors, and the representative color selection process from this hue range is completed. To do. However, if NO, the process proceeds to S903, the bin having the highest frequency within the hue range to be determined is adopted as the representative color, and the process proceeds to S904. Since the luminance level determination processing in S904 and S905 is synonymous with S702 and S703 shown in FIG. 7, description thereof is omitted here.

  In the first embodiment, the frequency maximum bin determination is performed only in the same chromatic hue range H1, but whether or not the frequency maximum bin is a representative color even when there is a bin only in the achromatic hue range H0. A determination may be performed. In this case, when the average of the hue values of all the bins is close to 0, the hue becomes indefinite. Therefore, when the average of the hue values of all the bins is close to the predetermined threshold value S ′ that separates the achromatic color region and the chromatic color region. carry out.

[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 (7)

By dividing a hue circle composed of saturation and hue based on a predetermined criterion, a plurality of regions in the hue ring are set, and further, a luminance direction of a predetermined color space is divided into a plurality of areas. A color space dividing means for setting a luminance level and defining a plurality of classes configured based on a plurality of regions set in the hue circle and a plurality of luminance levels set in the luminance direction;
For all the pixels included in the first image data, a frequency distribution is created by counting the number of pixels belonging to each class determined by the color space dividing means, and the colors of the pixels belonging to each class A frequency distribution creating means for obtaining an average value of information as color information of each class;
Determination means for determining whether or not a hue difference between classes is greater than or equal to a predetermined threshold based on the color information of each class obtained by the frequency distribution creating means;
If the determination means determines that the hue difference between the classes is greater than or equal to the predetermined threshold, the color information of the class having the maximum hue and the color information of the class having the minimum hue are selected as representative colors. When the determination means determines that the hue difference between the classes is not equal to or greater than the predetermined threshold, the frequency distribution created by the frequency distribution creation means is the largest luminance difference from the color information of the maximum class and the maximum class. Representative color selection means for selecting color information of a certain class as a representative color;
Replacement means for replacing each pixel value of a pixel included in the first image data using the selected representative color;
An image processing apparatus comprising: Input means for inputting image data;
Dividing means for dividing the image data input by the input means into a plurality of tile images,
The image processing apparatus according to claim 1, wherein the first image data is a tile image divided by the dividing unit. All of the classes in which the number of pixels is counted based on the frequency distribution created by the frequency distribution creating means belong to two or more of the plurality of areas set in the hue circle, or in one area Means for determining whether to belong;
If it is determined that the image belongs to the two or more areas, the color information of the class having the maximum frequency of the frequency distribution among the classes belonging to the area is used as a representative color, and further belongs to the area Means for selecting the color information of the class as a representative color if there is a class having a luminance difference of a certain level or more from the class having the highest frequency of the frequency distribution among the classes,
3. The image processing apparatus according to claim 1, wherein when it is determined that the image belongs to the one area, the determination unit and the representative color selection unit perform processing so as to select the representative color. 4. . All of the classes in which the number of pixels is counted based on the frequency distribution created by the frequency distribution creating means belong to two or more of the plurality of areas set in the hue circle, or in one area Means for determining whether to belong;
When determined to belong to the two or more areas, for each area,
Based on the color information of each class obtained by the frequency distribution creating means, it is determined whether the hue difference between classes is a predetermined threshold or more,
When it is determined that the hue difference between the classes is greater than or equal to the predetermined threshold, the color information of the class having the maximum hue and the color information of the class having the minimum hue are selected as representative colors,
If it is determined by the determination means that the hue difference between the classes is not greater than or equal to the predetermined threshold, the color information of the class having the maximum frequency of the frequency distribution among the classes belonging to the region is used as a representative color, Means for selecting the color information of the class as a representative color if there is a class having a luminance difference of a certain level or more from the class of the frequency distribution among the classes belonging to the region,
4. The process according to claim 1, wherein when it is determined that the image belongs to the one area, the determination unit and the representative color selection unit perform processing so as to select the representative color. 5. The image processing apparatus described.   The color space dividing unit sequentially scans pixels included in the first image data, and divides the hue circle using a pixel having a saturation equal to or higher than a predetermined threshold that is first found as a reference pixel. Setting a plurality of areas in the hue circle, further setting a plurality of luminance levels by dividing a luminance direction of a predetermined color space, and setting the plurality of areas set in the hue ring and the plurality of luminance directions 5. The image processing apparatus according to claim 1, wherein a plurality of classes configured based on the luminance level are determined. An image processing method of an image processing apparatus,
The color space dividing means sets a plurality of regions in the hue circle by dividing a hue circle constituted by saturation and hue based on a predetermined reference, and further sets a luminance direction of the predetermined color space. A color space dividing step of setting a plurality of luminance levels by dividing, and defining a plurality of classes configured based on a plurality of regions set in the hue circle and a plurality of luminance levels set in the luminance direction;
The frequency distribution creating means creates a frequency distribution by counting the number of pixels belonging to each class determined by the color space dividing means for all the pixels included in the first image data, A frequency distribution creating step for obtaining an average value of color information of pixels belonging to a class as color information of each class;
A determining step for determining whether or not a hue difference between the classes is equal to or greater than a predetermined threshold based on the color information of each class obtained in the frequency distribution creating step;
When the representative color selection unit determines that the hue difference between the classes is equal to or greater than the predetermined threshold in the determination step, the color information of the class having the maximum hue and the color information of the class having the minimum hue are obtained. While selecting as a representative color, if it is determined in the determining step that the hue difference between the classes is not greater than or equal to the predetermined threshold, the frequency information created by the frequency distribution creating means is the color information of the highest class and the maximum A representative color selection process for selecting, as a representative color, color information of a class having the most luminance difference from the class of
A replacement step of replacing each pixel value of the pixel included in the first image data by using the selected representative color;
An image processing method comprising:   A program for causing a computer to function as each unit of the image processing apparatus according to any one of claims 1 to 5.