JP2009260602A - Table data generator, and table data generation method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To easily and highly accurately generate a color mapping LUT for classifying images by a reference close to the color feeling of a human. <P>SOLUTION: The table data generator includes; a representative color acquisition means for acquiring the representative color of an image group from the image group for which images whose main color is one of selected colors selected for image classification beforehand are gathered; a histogram generation means for generating a color space histogram on the basis of the calculated representative color; a space value acquisition means for acquiring a color space value in the selected color from the color space histogram; and a table generation means for generating table data for the image classification using the color space value. It is preferable that the representative color acquisition means calculates the representative color from the respective images of the image group and defines the representative color of the calculated respective images as the representative color of the image group. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an apparatus for generating table data, such as a color mapping look-up table (LUT), and a method for generating table data, which are used when classifying images based on main colors.

The color mapping LUT is a table referred to for mapping an input color on a certain space. This color mapping LUT is generated by designating grid points in a discretized color space and tabulating color space data for colors corresponding to the designated grid points. When generating this color mapping LUT, a CIELAB color space, a CIEXYZ color space, or a CIECAM color space is used as a color space before conversion, and an HSV color space or an HSI color space is used as a color space after conversion. (See Patent Document 1). By using such a color space, the main colors used in the image can be optimally reproduced.
JP 2006-203896 A

  However, when using the YCbCr color space to create a color mapping LUT based on a standard close to human color space, the main color is selected in the HSV color space, and the selected main color is converted to the YCbCr color space. It is necessary to do work to do. This is a laborious work and is inefficient. In addition, since the main color is selected in the HSV color space, when an actual image is used, there is a difference between the impression of the color received from the image and the main color selected in the HSV color space. There is a problem that extraction accuracy is low.

  The present invention has been invented to solve the above-described problems, and can easily and accurately generate a color mapping LUT for classifying an image based on a standard close to human color sense. It is an object of the present invention to provide a table data generation apparatus and a table data generation method that can be performed.

  The table data generation device according to the first aspect of the present invention provides a representative color of an image group from an image group in which images having a color of the same system as the main color as a selected color selected in advance as a reference for classifying the image by color are collected. Representative color acquisition means for acquiring, histogram generation means for generating a color space histogram based on the representative color, space value acquisition means for acquiring a color space value of the basic color from the color space histogram, and the color space And a table generation means for generating table data for image classification using the value.

  In a second aspect based on the first aspect, the representative color acquisition unit calculates a representative color from each image in the image group, and sets the representative color of each acquired image as a representative color in the image group. It is characterized by.

  According to a third aspect, in the first and second aspects, the color space in the color space histogram is a color space indicated by hue, saturation, and lightness, and the color space in the table is luminance and chromaticity. It is a color space indicated by

  According to a fourth aspect of the present invention, in any one of the first to third aspects, the image group is a collection of images whose main colors are the same system color as the basic color. It is provided for each color.

  According to a fifth aspect of the present invention, there is provided a table data generation method, wherein a representative color of an image group is selected from an image group in which images having a color of the same system as a selected color preliminarily selected as a reference for classifying the image by color are collected. A representative color acquisition step of acquiring, a histogram generation step of generating a color space histogram based on the representative color, a space value calculation step of calculating a color space value of the basic color from the color space histogram, and the color space A table generation step of generating table data for image classification using the value.

  In a sixth aspect based on the fifth aspect, the representative color acquisition step acquires a representative color from each image of the image group, and sets the representative color of the acquired image as a representative color of the image group. It is characterized by.

  According to a seventh aspect, in the fifth or sixth aspect, the color space in the color space histogram is a color space indicated by hue, saturation, and brightness, and the color space in the table data is luminance and color. It is a color space expressed in degrees.

  In an eighth invention according to any one of the fifth to seventh inventions, the image group is a collection of images whose main colors are the same system color as the basic color. It is provided for each color

  According to the present invention, it is possible to efficiently and accurately generate a color mapping LUT used when classifying an image based on a reference close to a human color interval.

  FIG. 1 shows a schematic diagram of a computer used in generating a color mapping look-up table (LUT). The computer 10 includes a CPU 15, a ROM 16, a RAM 17, a storage 18, a mouse 19, a keyboard 20, and a display 21, and each unit is connected via a bus 22. The CPU 15 controls each unit of the computer 10 by executing a control program (not shown). The ROM 16 stores the control program described above, the LUT generation program 25 described later, and data used by these programs.

  The RAM 17 temporarily stores operators generated when the above-described programs are executed. Further, the RAM 17 is provided with a histogram storage area 26 for storing a histogram described later.

Examples of the storage 18 include an optical disk drive and a hard disk drive (HDD). The storage 18 includes an image storage area 30. In the image storage area 30, a plurality of folders for storing image data are provided. These folders are generated when the user operates the mouse 19 or the keyboard 20 or executes a program. Hereinafter, a case where four folders are provided in the image storage area 30 will be described. For example, the folder F1 with the folder name “image group 1” stores image data Pr _i (i = 1, 2,...) Whose main color is “red”, and the folder with the folder name “image group 2”. In F2, image data Pb _j (j = 1, 2,...) Whose main color is “blue” is stored. Further, the image data Pg _k (k = 1, 2,...) Whose main color is “green” is stored in the folder F3 of the folder name “image group 3”, and the folder name “image group 4”. the folder F4, "red", the image data _{Po l (l = 1,2, ···} ) which the the principal color color other than "blue" and "green" (other colors) are stored. For example, when the main color is “red”, the main color includes a red system color in addition to the red indicated by the three primary colors. In this way, the image data is stored for each folder, whereby the image data is classified (grouped) for each main color. Note that “red”, “blue”, “green”, and “other colors” described above are colors (hereinafter referred to as “selected colors”) that are selected in advance as a reference for classifying an image by color.

  The main colors used when classifying the image data are selected by the user or automatically when the above-described LUT generation program 25 is executed. The above-described red, blue, green, and others It is not limited to four types of colors. As a method for storing image data for each folder, the user estimates the main color of the image displayed on the display 21, and the image data that is the basis of the image is stored in the corresponding folder in the corresponding folder. Thus, the mouse 19 and the keyboard 20 are operated.

  The CPU 15 causes the computer 10 to function as a table data generation device by executing the LUT generation program 25 stored in the ROM 16. As shown in FIG. 2, when the CPU 15 executes the LUT generation program 25, the CPU 15 has functions of a representative color acquisition unit 35, a histogram generation unit 36, an HSV calculation unit 37, and an LUT generation unit 38.

  The representative color acquisition unit 35 acquires a representative color for each folder using image data stored in each folder in the image storage area 30 of the storage 18. As a specific procedure for acquiring the representative color, the representative color of the image data is acquired for all the image data stored in the same folder, and the representative color of all the obtained image data is set as the representative color of the folder. To do.

  For example, when the color information of the image data includes color space values of hue (H) value, saturation (S) value, and lightness (V) value, the representative color acquisition unit 35 first determines the hue value from the color information of each pixel. And a histogram for the hue value (hereinafter referred to as H histogram) is generated in the histogram storage area 26 of the RAM 17.

  FIG. 3 shows an image P whose main color is blue and an H histogram obtained from this image. An image P shown in FIG. 3A is an image in which a flag with a blue base and a white cross line and a blue sky are reflected, and the main color of the image P is blue. FIG. 3B shows an H histogram of the image shown in FIG. 3A, where the horizontal axis represents the hue value and the vertical axis represents the frequency. The hues corresponding to the hue values are in the order of “red”, “yellow”, “green”, “blue”, “purple”. When the H histogram in the image P is generated, the frequency of the hue value of “blue” becomes higher than the frequency of the hue value of other colors.

  FIG. 3C is a diagram in which the hue value of “blue” is cut out from the H histogram of FIG. The representative color acquisition unit 35 extracts a hue value having a high frequency in the H histogram as a process of acquiring the representative color. For example, in the H histogram shown in FIG. 3B, since the frequency with respect to the hue value of “blue” is high, the hue values at points D1 to D10 having high frequencies in the blue hue value are extracted. After the hue value is extracted, the saturation value corresponding to the extracted hue value and the lightness value are extracted for each hue value using the color information of the image data that is the basis of the image P. Thereby, the representative color of the image is acquired. The representative color extraction in the representative color acquisition unit 35 is performed for all image data stored in the same folder and for all folders.

  The histogram generation unit 36 generates an H histogram, a histogram for saturation values (S histogram), and a histogram for lightness values (V histogram) using the representative colors extracted by the representative color acquisition unit 35 (see FIG. 4). ).

  The HSV calculation unit 37 refers to the H histogram, S histogram, and V histogram created by the histogram generation unit 36, and generates color information from a range that the color space value of the hue value, saturation value, and brightness value can take. . As a procedure for generating this color information, a range that can be taken for each color space value and a minimum value and a maximum value for each color space value are obtained from each histogram. Since these histograms are generated from representative colors acquired from each image, the variation is large when the representative colors are small, and the bias is large when the representative colors are large. Regarding the hue value, if the range that the hue value can take is known, the color information using the hue value within the range becomes color information indicating an approximate color. For this reason, in each histogram, if the range that the color space value can take is known, a value obtained by equally dividing the possible range is virtually obtained and complemented, and these values are combined to combine the colors included in the selected color. Information can be generated without variation, and a color mapping LUT can be generated with high accuracy.

  The LUT generation unit 38 generates a color mapping LUT 40 using the color information acquired by the HSV calculation unit 37. As described above, the generated color information uses color space values indicated on the HSV color space, whereas the color information of image data acquired by imaging is indicated on the YCbCr color space. Color space values are used. For this reason, the acquired color information is converted from a color space value composed of a hue value, a saturation value, and a lightness value into a color space value composed of a Y value, a Cb value, and a Cr value. That is, the color information indicated in the HSV color space is converted into color information indicated in the YCbCr color space. This conversion is performed using the following equation.

  The color space value thus coordinate-transformed is plotted on the Y, Cb, and Cr coordinates. As shown in FIG. 5, the Y, Cb, and Cr coordinates are equally divided in the Y direction by, for example, the number of divisions 8, and each of the Cb and Cr directions has a dense central portion and a central portion to a peripheral portion. It is divided into 8 so as to become sparse toward. In the present embodiment, the Cb direction and the Cr direction are each divided so as to have a ratio of 4: 2: 1: 1: 1: 1: 2: 4. When the coordinate-converted color space value is plotted on the divided Y, Cb, and Cr coordinates, it corresponds to one of the Y, Cb, and Cr coordinate areas. If the color space value after coordinate conversion is obtained from image data classified as “blue” among the selected colors, “blue” is assigned to the corresponding area when plotted. In this way, one of the selected colors is assigned to the region corresponding to the plot of all the color space values that have undergone coordinate conversion, and the color mapping LUT 40 is generated. When an area obtained by dividing the Y, Cb, and Cr coordinates is assigned as one of the selected colors, the assigned area is labeled according to the selected color. For example, “1” is labeled when assigned to “red”, and “2” is labeled when assigned to “blue”. Further, “3” is labeled when assigned to “green”, and “4” is labeled when assigned to “other colors”.

  Next, a procedure for generating the color mapping LUT 40 will be described with reference to the flowchart of FIG. When the color mapping LUT 40 is generated, image data classified by the user for each selected color is stored in advance in the image storage area 30 of the storage 18 for each folder. Step S101 is processing for initializing the histogram storage area 26 of the RAM 17.

  Step S102 is processing for extracting a representative color of the image data in the folder. The processing in step S102 is executed by the representative color acquisition unit 35. First, the image data Pr1 stored in the folder F1 with the folder name “image group 1” is read, and an H histogram is generated in the histogram storage area 26 of the RAM 17. From the generated H histogram, points with higher frequencies are extracted from, for example, 10 locations (D1 to D10 in FIG. 3) from the higher frequency, and the hue values for the extracted points with high frequency are read out. The representative color is extracted by acquiring the saturation value and the lightness value corresponding to the read hue value using the image data Pr1.

  Step S103 is processing for determining whether or not the representative colors for the image data stored in the same folder have been extracted from all the image data. If it is determined in this determination that the representative colors have been extracted from all the image data (if the determination in step S103 is Yes), the process proceeds to step S104. On the other hand, when it is determined that the extraction of the representative color has not been executed for all the image data (No in the determination in step S103), the process returns to step S102.

  Step S104 is a process of generating an H histogram, an S histogram, and a V histogram based on the extracted representative colors. The processing in step S104 is executed by the histogram generation unit. In step S102, since representative colors are extracted from all the image data in the folder, the representative colors of all the extracted image data are used in the process of step S104. For example, when the image data stored in the folder F1 is Pr1 to Pr3, 30 representative colors are extracted from each image data, a total of 30 representative colors. A histogram, an S histogram, and a V histogram are generated, respectively.

  Step S105 is processing for obtaining color information from the H histogram, S histogram, and V histogram generated by the processing of step S104. This process is executed by the HSV calculation unit 37. First, in the H histogram, the range that the hue value can take is obtained, the minimum hue value and the maximum hue value of the possible range are obtained, and then each hue value when the possible range is equally divided is obtained. calculate. Similarly, using the S histogram and the V histogram, the saturation value, the minimum value, and the maximum value of the saturation value when the minimum value, the maximum value, and the range that the saturation value can take are equally divided. And each brightness value when the range which the brightness value can take is equally divided is calculated. Then, the color information is generated by associating the saturation value and the lightness value corresponding to each of the hue values. Thereby, the color information for each folder, that is, the color information for the selected color is acquired.

  Step S106 is a step of determining whether or not acquisition of color information has been completed for all folders. If color information has been acquired for all folders (Yes in step S106), the process proceeds to step S107. On the other hand, if color information has not been acquired for all folders, the process proceeds to step S101. Thereby, the color information for each selected color classified by the user is acquired for each selected color.

  Step S107 is processing to generate the color mapping LUT 40 using the color information acquired for each folder. The processing in step S107 is executed by the LUT generation unit 38. First, the color information shown in the HSV color space generated in step S106 is converted into color information shown in the YCbCr color space using the above-described formula. Since the color information generated in step S106 includes color space values of hue value, saturation value, and brightness value, these color space values are converted into color space values of saturation value and chromaticity value. This coordinate transformation is executed for each folder.

  After coordinate-converting the color information of all folders, plot it in the Y, Cb, and Cr coordinates. Since the Y, Cb, and Cr coordinates are divided into a plurality of areas in advance, when the color information subjected to coordinate conversion is plotted on the Y, Cb, and Cr coordinates, one of the plurality of areas described above corresponds. This corresponding area is an area assigned to the selected color. That is, a color mapping LUT is generated by assigning areas using the color information of all folders. Note that the assigned area is labeled with the selected color.

  In this embodiment, without using the hue value, saturation value, and lightness value of the representative color extracted from the H histogram, the color information is obtained by combining the color space values when the range that these color space values can take is equally divided. However, it is not necessary to limit to this. Since these histograms use discrete data, the interval between adjacent color space values is widened in the histogram. When the interval is wider than the other intervals, the wide range can be equally divided, and the color space value can be virtually calculated before generating color information.

  In this embodiment, the color mapping LUT is generated on the computer. However, the present invention is not limited to this, and it is also possible to generate the color mapping LUT in the digital camera.

It is a functional block diagram which shows the electric constitution of a computer. It is a functional block diagram which shows the structure of CPU. It is a figure which shows an example of an image and H histogram in an image. It is a figure which shows an example of H histogram, S histogram, and V histogram. It is a figure which shows an example at the time of dividing | segmenting a YCbCr coordinate into a some area | region. It is a follow chart which shows the flow which produces | generates the LUT for color mapping.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... Computer, 15 ... CPU, 16 ... ROM, 17 ... RAM, 18 ... Storage, 25 ... LUT generation program, 26 ... Histogram storage area, 35 ... Representative color acquisition part, 36 ... Histogram generation part, 37 ... HSV calculation part , 38 ... LUT generation unit, 40 ... LUT for color mapping

Claims (8)

Representative color acquisition means for acquiring a representative color of the image group from an image group in which images having the same color as the main color as the selected color selected as a reference for classifying the image by color are collected;
Histogram generating means for generating a color space histogram based on the representative color;
Space value acquisition means for acquiring a color space value in the selected color from the color space histogram;
Table generating means for generating table data for image classification using the color space value;
A table data generation device comprising: The table data generation device according to claim 1,
The table color generation device, wherein the representative color acquisition unit acquires a representative color from each image of the image group, and sets the representative color of each acquired image as the representative color of the image group. In the table data generation device according to claim 1 or 2,
The color space in the color space histogram is a color space indicated by hue, saturation, and lightness, and the color space in the table is a color space indicated by luminance and chromaticity. apparatus. In the table data generation device according to any one of claims 1 to 3,
The table group is a table data generation device characterized in that the image group is a collection of images whose main colors are the same system color as the selected color, and the image group is provided for each selected color. A representative color calculation step of obtaining a representative color of the image group from an image group in which images having the same system color as the selected color selected as a reference for classifying the image by color are collected;
A histogram generation step of generating a color space histogram based on the representative color;
A space value calculating step of calculating a color space value in the selected color from the color space histogram;
A table generation step of generating table data for image classification using the color space value;
A table data generation method characterized by comprising: In the table data generation device according to claim 5,
The representative color acquisition step acquires a representative color from each image of the image group, and uses the acquired representative color of each image as a representative color of the image group. In the table data generation method according to claim 5 or 6,
The color space in the color space histogram is a color space indicated by hue, saturation, and brightness, and the color space in the table data is a color space indicated by luminance and chromaticity. Generation method. In the table data generation method according to any one of claims 5 to 7,
The table group is a table data generation method characterized in that images having the same system color as the selected color as a main color are collected, and the image group is provided for each selected color.

Images