JP2008245116A - Image processing apparatus, image processing method, and image processing program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve a problem wherein versatility of statistical data is low and a processing amount for obtaining the statistical data is large. <P>SOLUTION: An image processor includes a statistic generating means for acquiring input image data represented in an equipment-dependent color space and generates a predetermined statistic from the input image data while the color space is maintained, and a statistic converting means for converting the generated statistic into color information represented in the equipment-dependent color space. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an image processing apparatus, an image processing method, and an image processing program.

  Conventionally, as one type of image processing, processing for generating a statistical value of input image data based on the value of each pixel of the input image data has been performed. Further, a process of obtaining a correction amount for the input image data based on the statistical value, correcting a value of each pixel of the image data according to the obtained correction amount, and passing the corrected image data to the next processing step. It is done.

As a specific example, among the pixels of the input image data, the pixels corresponding to a specific memory color (for example, skin color) are totaled, and an average value (statistical value) for each of RGB of the totaled pixels is obtained. A correction amount ΔR, ΔG, ΔB (correction amount) for each of RGB is obtained according to a difference between the value and an average value (target value) for each ideal RGB of the skin color pixels obtained in advance, and this correction amount There is known a color correction device that generates tone curves for each of RGB in accordance with ΔR, ΔG, and ΔB, converts the element colors of all the pixels of the image data by using the tone curves, and corrects the image data ( (See Patent Document 1).
Japanese Patent Laid-Open No. 11-8773

Here, the process up to the generation of the statistical value and the calculation and correction process of the correction amount using the statistical value are basically separated processes. Therefore, once the statistical value is generated, it is possible to repeatedly perform the correction process while changing the calculation amount of the correction amount and the like while using the statistical value. That is, the above-mentioned statistical amount is a numerical value assumed to be reused due to its character.
However, when the statistical value is generated based on the input image data, the following problem has occurred. In image processing, the color space adopted by input image data may be different from the color space adopted as a work color space for calculation of correction amounts and correction processing. For example, the input image data is data expressed in the sRGB color space, but the work color space is set to a color space having a color reproduction range different from the sRGB color space. Therefore, if the statistical value is information expressed in the color space adopted by the input image data, the statistical value may not be used for calculation of the correction amount as it is due to the difference in the color space.

Therefore, in the prior art, first, the value of each pixel of the input image data is converted into a value in the working color space, and the statistical value is calculated based on the input image data after the color space conversion, whereby the difference in the color space described above is obtained. The inconvenience of using statistical values due to the problem was solved.
However, converting the input image data to the color space before generating the statistical value significantly increases the amount of calculation processing, and increases the time required for image processing.
In addition, the above document does not assume the difference between the color space of the input image data and the work color space, so there is no contrivance in terms of ease of use of statistical values.

  The present invention has been made in view of the above problems, and provides an image processing apparatus, an image processing method, and an image processing that achieves both improvement in versatility of statistical values used for correcting image data and reduction in processing time. The purpose is to provide a program.

  In order to achieve the above object, in the image processing apparatus according to the present invention, the statistical value generation means acquires the input image data expressed by the device-dependent color space and maintains the color space from the input image data. Generate statistics for. Then, the statistical value conversion means converts the statistical value into color information expressed by a device-independent color space. As the statistical value referred to here, various values such as an average value, a maximum value, a minimum value, a median value, and a mode value are assumed. In other words, according to the present invention, the statistical value is obtained as information on the device-independent color space, so that the versatility is very high. Since it is not a statistical value adapted to only one work color space as in the past, any work space can be converted into information suitable for each work space to calculate the correction amount. Can be used. Further, unlike the prior art, since the input image data is not converted to a value in the working color space before the statistical value is generated, the processing amount required for generating the statistical value is greatly reduced.

  Here, the image processing apparatus converts a specific color gamut defined in the device-independent color space into a color gamut dependent on the color space of the input image data, and after the conversion, among the pixels constituting the input image data A pixel extraction unit that extracts pixels belonging to the color gamut may be further provided. The statistical value generation means may generate a statistical value based on the extracted pixel group. With such a configuration, it is possible to accurately generate statistical values for pixels belonging to a specific color gamut among the pixels of the input image data regardless of the color space of the input image data, and the statistical values are highly versatile. It can be obtained as information on a device-independent color space.

The statistical value generating means may divide the input image data into a plurality of regions and generate a statistical value for each divided region. With such a configuration, the statistical value generated for each area of the input image data can be obtained as information on a highly versatile device-independent color space.
Further, the statistical value generation means may detect a human face existing in the input image data and generate a statistical value based on a pixel group in the detected human face region. With this configuration, it is possible to obtain statistical values generated for the human face in the input image data as information on a highly versatile device-independent color space.

  Further, the statistical value generation means may generate a statistical value for each element color constituting the input image data. For example, when each pixel of the input image data is composed of each element color of RGB, a statistical value is generated for each RGB. The statistical value conversion means regards a combination of statistical values of the same type for each element color as one color, and converts the combination into color information expressed by a device-independent color space. The combination of statistical values of the same type for each element color is, for example, a combination of the maximum value of R, the maximum value of G, and the maximum value of B. In general, a combination of statistical values of the same type for each element color does not necessarily have a pixel with this combination, but each statistical value for one combination belongs to a pixel whose color is close to each other. Is most. Therefore, even if a combination of the same kind of statistical values for each element color is regarded as one color and converted into color information in the device-independent color space, it can be said that there is very little deterioration in the statistical value. Further, according to this configuration, when various types of statistical values are generated for each element color of the input image data, each statistical value can be efficiently converted into the device-independent color space.

  The statistical value generation means acquires a plurality of input image data having a common image type, generates a statistical value from each of the plurality of input image data, averages the statistical value for each of the plurality of input image data, The statistical value conversion means converts the averaged statistical value into color information expressed by a device-independent color space, and stores the converted color information as a statistical value for the common image type as a predetermined value. It may be stored in the area. The stored statistical value can be said to be a statistical value generally applicable to the common image type (for example, landscape image). Therefore, after saving the statistical value, when trying to correct new input image data related to the common image type, it is not necessary to generate a statistical value from the new image. A correction amount can be calculated and corrected using the value.

  Up to now, the technical idea according to the present invention has been described in the category of image processing apparatus. However, the invention of the image processing method including the processing steps respectively corresponding to the respective units included in the image processing apparatus, and the image processing apparatus It goes without saying that the invention of an image processing program for causing a computer to execute a function corresponding to each means included in can be grasped.

The embodiment of the present invention will be described in the following order.
(1) Schematic configuration of image processing apparatus, etc. (2) Memory color pixel extraction process (3) Statistical data acquisition process (4) Correction amount calculation and correction process (5) Modification (6) Summary

(1) Schematic Configuration of Image Processing Device, etc. FIG. 1 is a block diagram showing a schematic configuration of an image processing device and peripheral devices according to an embodiment of the present invention. FIG. 1 shows a computer 20 as an image processing apparatus that plays a central role in image processing. The computer 20 includes a CPU 21, a ROM 22, a RAM 23, a hard disk (HD) 24, and the like. The computer 20 is appropriately connected with a scanner 11 as various image input devices, a digital still camera 12, a printer 31 as various image output devices, and a display 32.

  In the computer 20, the CPU 21 executes various programs stored in a predetermined storage medium such as the ROM 22 or the HD 24 while using the RAM 23 as a work area. In the present embodiment, the CPU 21 reads and executes an application (APL) 25 stored in the HD 24. The APL 25 to be executed generally constitutes functional blocks such as an image data acquisition unit 25a, a pixel extraction unit 25b, a statistical data acquisition unit 25c, a correction amount acquisition unit 25d, and a correction unit 25e.

The image data acquisition unit 25a acquires image data output from the image input device or image data stored in advance in the HD 24 as input image data.
The pixel extraction unit 25b extracts (samples) pixels that satisfy a predetermined condition from the pixels constituting the input image data. In the present embodiment, pixels corresponding to the color gamut of a specific memory color are extracted. The specific memory color means a specific color that is particularly important in the image, such as skin color, sky blue, green, and red.

  The statistical data acquisition unit 25c generates various statistical values (statistical data) based on the pixel group extracted by the pixel extraction unit 25b, the pixel group of the input image data passed from the image data acquisition unit 25a, and the like. get. The statistical data acquisition unit 25c generates a histogram based on the pixel group, and generates statistical data from the aggregation result of the histogram. The statistical data generation unit 25c1 uses the generated statistical data as a predetermined device-independent color space. And a statistical data conversion unit 25c2 that stores the converted color information in the HD 24 as statistical data 24c.

  The correction amount acquisition unit 25d reads the statistical data 24c from the HD 24, acquires the correction target value 24b, and determines a correction amount for correcting the input image data. The correction target value 24b is an ideal value of the statistical data 24c and is stored in advance in a predetermined storage medium such as the HD 24 or input from the outside. In the present embodiment, the correction target value 24b is information defined in a predetermined device-independent color space. The correction amount acquisition unit 25d includes a target value conversion unit 25d1 that converts the correction target value 24b into information in a predetermined output color space, and a statistical data conversion unit 25d2 that converts statistical data 24c into information in the predetermined output color space. The correction amount determination unit 25d3 calculates a correction amount based on a comparison between the converted statistical data and the correction target value.

The correction unit 25e receives the input image data from the image data acquisition unit 25a, corrects the input image data in units of pixels based on the determined correction amount, and outputs the corrected image data.
Details of processing by the pixel extraction unit 25b, the statistical data acquisition unit 25c, the correction amount acquisition unit 25d, and the correction unit 25e will be described later.

  In the computer 20, the APL 25 is incorporated into the operating system 26 and executed. The operating system 26 also incorporates a printer driver 27 and a display driver 28. The display driver 28 is a driver that controls image display on the display 32, and can display an image on the display 32 based on the corrected image data output from the correction unit 25 e of the APL 25. In addition, the printer driver 27 performs color conversion processing and halftone processing on the ink color (for example, cyan, magenta, yellow, black) color system for the corrected image data output from the correction unit 25e of the APL 25. It is possible to cause the printer 31 to print an image based on the print data by generating print data by executing a rasterization process or the like, and outputting the print data to the printer 31.

  Here, the input image data acquired by the image data acquisition unit 25a is a dot matrix in which each element color of RGB is expressed by a plurality of gradations (for example, 256 gradations from 0 to 255) to define the color of each pixel. Data. The RGB color space is generally the sRGB color space, but there is a color space having a color reproduction range wider than the sRGB color space, such as the AdobeRGB (Adobe is a registered trademark of Adobe systems) color space. Accordingly, a color space (input color space) adopted by input image data output from an image input device such as the digital still camera 12 may be an sRGB color space or a color space different from the sRGB color space. On the other hand, the setting of the color space (output color space) for expressing the image data output from the correction unit 25e, that is, the image data that the APL 25 passes to the next processing step such as the printer driver 27 is the same as the input color space. Sometimes it is different.

  In this embodiment, a configuration capable of executing appropriate correction on input image data regardless of what the input color space is and whether the input color space and the output color space are the same is realized. . In the following, for ease of explanation, each of the input color space and the output color space is either a sRGB color space or a specific color space (wRGB color space) having a wider color reproduction range than the sRGB color space. And Both the sRGB color space and the wRGB color space are device-dependent color spaces.

  Note that all or a part of the processing executed by the computer 20 described in the present embodiment may be executed on the image input device side, or may be executed on the image output device side. For example, the function of the correction unit 25e is provided on the printer 31 or display 32 side, and the printer 31 or display 32 corrects the image data transferred from the APL 25 and prints based on the corrected image data. Processing or image display processing may be performed. In this case, it can be said that an image processing system is constructed by the computer 20, the printer 31, and the display 32.

(2) Memory Color Pixel Extraction Processing FIG. 2 is a part of image processing executed by the computer 20 according to the APL 25, and mainly shows processing executed according to the pixel extraction unit 25b in a flowchart.
In step S (hereinafter, step description is omitted) 100, the computer 20 determines in which color space the input image data is represented. The input image data acquired by the image data acquisition unit 25a is added with a header describing various attached information related to the image data, and information indicating the type of the input color space is described in the header. The computer 20 determines the type of the input color space based on the information described in this header.

In S110, the computer 20 reads the color gamut definition information 24a stored in advance in the HD 24. However, the computer 20 may input the color gamut definition information 24a from the outside of the computer 20. The color gamut definition information 24a is information defining a memory color range in the device-independent color space. In this embodiment, the L ^* a ^* b ^* color space defined by the International Commission on Illumination (CIE) is adopted as the device-independent color space. Accordingly, the color gamut definition information 24a defines each range of lightness (L ^* ), saturation (C ^* ), and hue (h ^* ) for a memory color (hereinafter, “ ^* ” is omitted).

FIG. 3 shows an example of the color gamut A defined by the color gamut definition information 24a of a certain memory color (for example, skin color) in the Lab color space. As shown in the figure, the color gamut A has ranges of lightness L, saturation C, and hue h,
L _s ≦ L ≦ L _e
C _s ≦ C ≦ C _e
h _s ≤ h ≤ h _e
It is expressed as a solid divided by. In the figure, the projection view of the color gamut A is also shown hatched on the ab plane.
In S120, the computer 20 sets a conversion reference point P on each surface of the color gamut A. The conversion reference point P means a coordinate value to be converted into information depending on the color space of the input image data among the coordinates on the outline of the color gamut A.

  As shown in FIG. 4, the computer 20 sets conversion reference points P (P1 to P6) one by one on each surface of the color gamut A. Various ideas can be adopted as to where the position of the conversion reference point P is on each surface, but in this embodiment, the center positions of each surface are set as conversion reference points P1 to P6, respectively. Here, the conversion reference point P1 is a point on the surface having the smallest hue angle in the color gamut A, and the conversion reference point P2 is a point on the surface having the largest hue angle in the color gamut A. The reference point P3 is a point on the surface with the lowest saturation in the color gamut A, the conversion reference point P4 is a point on the surface with the highest saturation in the color gamut A, and the conversion reference point P5 is , The point on the surface having the lowest lightness in the color gamut A, and the conversion reference point P6 is the point on the surface having the highest lightness in the color gamut A.

  In S130, the computer 20 converts the coordinate data Lab of the conversion reference point P into information that depends on the input color space. First, the computer 20 acquires an ICC profile corresponding to the type of the input color space determined in S100 from a predetermined storage medium such as the HD 24, and refers to the ICC profile to obtain the coordinate data Lab of each conversion reference point P. Convert to RGB data. When the input color space is the sRGB color space, an ICC profile that defines the conversion correspondence between the Lab data and the RGB data in the sRGB color space is obtained for a plurality of reference points in the Lab color space, and the profile is referred to By doing so, the Lab data of each conversion reference point P is converted into RGB data.

  When it is said that the coordinate value of the conversion reference point P is converted into information dependent on the input image color space, the simplest is information in which the RGB data itself obtained by conversion as described above depends on the input color space. It corresponds to. However, in the present embodiment, the RGB data thus converted is further converted into HSV format information. H (Hue) means hue, S (Saturation) means saturation, and V (Value) means lightness. Conversion from RGB to HSV can be executed by a known conversion formula.

As a result of such processing, the conversion reference points P1 to P6 shown in FIG.
_{P1 (L 1, a 1,} b 1) → P1' (H 1, S 1, V 1)
_{P2 (L 2, a 2,} b 2) → P2' (H 2, S 2, V 2)
_{P3 (L 3, a 3,} b 3) → P3' (H 3, S 3, V 3)
_{P4 (L 4, a 4,} b 4) → P4' (H 4, S 4, V 4)
_{P5 (L 5, a 5,} b 5) → P5' (H 5, S 5, V 5)
_{P6 (L 6, a 6,} b 6) → P6' (H 6, S 6, V 6)
It was converted as follows.

  Since such HSV values are values determined by the RGB values of the conversion source, it can be said that the information depends on the input color space. Of course, when the input color space is the wRGB color space, the computer 20 refers to the ICC profile that defines the conversion correspondence from the Lab color space to the wRGB color space, and converts the Lab data of each conversion reference point P to the RGB data. The converted RGB data is expressed in the HSV format.

  The computer 20 does not perform the two-step conversion process of converting Lab data into RGB data and converting it into HSV data as described above. Instead, the computer 20 performs Lab data and HSV data for each type of input color space. A look-up table (LUT) that defines the conversion correspondences may be generated in advance and stored in a predetermined storage medium such as the HD 24. That is, each RGB data in the ICC profile that defines the conversion correspondence from the Lab color space to the sRGB color space (or wRGB color space) is converted into HSV data, and each HSV data obtained by this conversion is converted into the conversion source By associating with Lab data corresponding to RGB data, an LUT that can convert Lab data into HSV data depending on the sRGB color space (or wRGB color space) is generated.

  In S <b> 130, the computer 20 may convert the Lab data of each conversion reference point P into HSV data by selecting and referring to each LUT generated in this way according to the type of the input color space.

In S140, the computer 20 determines a color gamut for pixel extraction based on the information dependent on the input color space obtained by the conversion process in S130.
In the present embodiment, since the coordinate values P1 ′ to P6 ′ are obtained as the conversion results of the conversion reference points P1 to P6, the computer 20 uses the color values for pixel extraction based on these coordinate values P1 ′ to P6 ′. The area A ′ is determined. The coordinate value P1 ′ is a result of converting the conversion reference point P1 on the surface having the smallest hue angle in the color gamut A, and the coordinate value P2 ′ is the conversion reference on the surface having the largest hue angle in the color gamut A. since the result of converting the point P2, the range sandwiched between the hue of H ₂ hue H ₁ and the coordinate values of the coordinate values P1'P2' hue range of the color gamut A'.

The coordinate value P3 ′ is the result of converting the conversion reference point P3 on the surface with the lowest saturation in the color gamut A, and the coordinate value P4 ′ is the conversion on the surface with the highest saturation in the color gamut A. since the result of converting the reference point P4, the range sandwiched between the saturation S ₃ and the coordinate values saturation S ₄ of P4' coordinates P3' the saturation range of the color gamut A'.
The coordinate value P5 ′ is a result of converting the conversion reference point P5 on the surface having the lowest lightness in the color gamut A, and the coordinate value P6 ′ is the conversion reference point on the surface having the highest lightness in the color gamut A. Since it is the result of converting P6, the range between the brightness V ₅ of the coordinate value P5 ′ and the brightness V ₆ of the coordinate value P6 ′ is set as the brightness range of the color gamut A ′.

FIG. 5 shows the color gamut A ′ in the HSV color space. As shown in the figure, the color gamut A ′ is
H ₁ ≦ H ≦ H ₂
S ₃ ≦ S ≦ S ₄
V ₅ ≦ V ≦ V ₆
It is expressed as a solid divided by.
As described above, the color gamut (color gamut A) defined by the color gamut definition information 24a can be converted into the color gamut (color gamut A ′) depending on the color space of the input image data by the processing of S100 to S140. Also, by defining the color gamut after conversion based on each range of lightness, saturation, and hue, the similarity in shape between the color gamut before conversion and the color gamut after conversion can be maintained.

In S150, the computer 20 extracts pixels belonging to the converted color gamut from the pixels constituting the input image data. In this case, after expressing the RGB data of each pixel as HSV data, it is determined whether or not it belongs to the color gamut after the conversion, and only the pixel to which it belongs is extracted. The extracted pixel group is transferred to the above-described statistical data acquisition unit 25c, and becomes data from which the statistical data 24c is calculated.
3 and 4, only the color gamut relating to one color gamut definition information 24a is shown, but the color gamut definition information 24a exists for each kind of memory color as described above. In the present embodiment, it is assumed that there is color gamut definition information 24a in which color gamuts of at least skin color, sky blue, green, red, and low chroma are defined. The computer 20 converts each color gamut into a color gamut depending on the input color space for each color gamut definition information 24a, and performs pixel extraction for each color gamut after conversion.

(3) Statistical Data Acquisition Processing FIG. 6 is a part of image processing executed by the computer 20 according to the APL 25, and mainly shows processing executed according to the statistical data acquisition unit 25c in a flowchart.
In S200, the computer 20 generates statistical data for the pixel group extracted for each memory color as described above. Specifically, a histogram for each element color RGB is generated based on a pixel group related to one memory color, and average values Rav, Gav, and Bav in the respective histograms for each RGB are calculated.

FIGS. 7A to 7C illustrate RGB histograms. Each histogram is a frequency distribution in which the number of pixels is defined on the vertical axis and the gradation value (0 to 255) is defined on the horizontal axis.
The average values Rav, Gav, and Bav are calculated for each type of memory color. Therefore, in S200, an average value for each memory color and for each RGB is generated for the input image data. However, statistical data is not generated for memory colors in which the number of pixels extracted in S150 does not exist over a predetermined ratio with respect to the total number of pixels of the input image data. This is because it is less necessary to perform correction for colors that are not present in the input image data or are very few even if they exist.

  In S210, the computer 20 generates statistical data for the entire input image data. Specifically, a histogram for each RGB is generated for pixels in the entire range of the input image data, and the maximum values Rmax, Gmax, and Bmax and the minimum values Rmin, Gmin, and Bmin in each histogram for each RGB are acquired. To do. When obtaining the maximum value and the minimum value, the gradation value on the highest gradation side on the histogram may be simply set as the maximum value, and the gradation value on the lowest gradation side may be set as the minimum value. However, in the present embodiment, the histograms are divided by a certain distribution ratio (for example, 0.5% of the number of images used for aggregation of the histogram) from the gradation value on the highest gradation side and the gradation value on the lowest gradation side. The positions inside are set to the maximum and minimum values, respectively. If the upper and lower edges of the histogram are cut by a predetermined ratio in this way, statistical data can be generated ignoring the effects of white spots and black spots caused by noise on the high gradation side and low gradation side. it can. Such cutting of the upper end and the lower end is also executed when generating statistical data based on another histogram.

In S220, the computer 20 divides the input image data into a plurality of areas and generates statistical data for each divided area.
FIG. 8 shows an example of a state in which the input image data D is divided into a plurality of regions. In the figure, the input image data D is divided into 5 parts vertically and horizontally to form a total of 25 regions T. Of course, the method of dividing the input image data D is not limited to the mode shown in FIG. The computer 20 generates a histogram for each RGB in units of divided areas, and calculates average values Rav, Gav, and Bav for each histogram for each RGB. Accordingly, in S220, an average value for each divided region and for each RGB is generated for the input image data. In S220, median (median) Rmed, Gmed, and Bmed of each histogram may be generated in addition to the average value.

In S230, the computer 20 detects a human face (human face) present in the input image data, and generates statistical data for a pixel group in the detected human face region. The human face detection process can be executed using a known human face detection program. Here, a histogram for each RGB is generated based on the detected pixel group within the range of the human face, and average values Rav, Gav, and Bav in each of the histograms for each RGB are calculated. When a plurality of human faces are detected, average values Rav, Gav, and Bav are calculated for each detected human face. Of course, if no human face is detected, statistical data about the human face is not generated.
In addition, when generating each of the above histograms, it may be generated based on all pixels belonging to the area to be aggregated of the histogram, but pixels extracted based on a predetermined extraction rate from the area to be aggregated You may generate based on.

  In S240, the computer 20 converts each piece of statistical data generated as described above into color information in the Lab color space. In other words, since each statistical data generated in S200 to S230 is information dependent on the input color space, it is converted into information on the Lab color space, which is a device-independent color space, for later calculation of correction amount. Statistical data can be easily used regardless of the color space (output color space) used. Specifically, in S240, a conversion profile (also referred to as an input device profile) corresponding to the input color space determined in S100 is obtained from a predetermined storage medium such as the HD 24, and the statistical data is referred to by referring to the conversion profile. Are converted into Lab data. When the input color space is the sRGB color space, a conversion profile for converting the sRGB color space to the Lab color space is used. When the input color space is the wRGB color space, the wRGB color space is changed to the Lab color space. Use a conversion profile to convert.

FIG. 9 shows an example in which each statistical data as information depending on the input color space is converted into Lab data. In the figure, the input color space is an sRGB color space. In FIG. 11 and FIG. 11 described later, Lab data is also expressed in 256 gradations from 0 to 255.
In the present embodiment, a combination of statistical data of the element colors RGB and the same kind of statistical data is regarded as one color, and this combination is converted into Lab data. That is, RGB data composed of a combination of maximum values Rmax, Gmax, Bmax of the histogram of the entire input image data, RGB data composed of average values Rav, Gav, Bav for a certain memory color, and an average for a certain divided area RGB data composed of values Rav, Gav, Bav, and RGB data composed of average values Rav, Gav, Bav relating to human face colors are converted into Lab data.

  The individual values (for example, the maximum values Rmax, Gmax, Bmax of the histogram) in the combination of the same kind of statistical data are not necessarily said to belong to the same pixel, but the colors are close to each other. It can be said that it belongs to a pixel. For example, it is assumed that the maximum values Rmax, Gmax, and Bmax all belong to whitish pixels, and the minimum values Rmin, Gmin, and Bmin all belong to blackish pixels. Also, the average values Rav, Gav, Bav for one memory color and the average values Rav, Gav, Bav for one divided area belong to pixels having similar colors. It is assumed that Furthermore, as described above, when obtaining statistical data from each histogram, the upper end and the lower end of the histogram are obtained after being cut by a predetermined distribution ratio, thereby preventing the influence of noise from appearing in the statistical data. . Therefore, it can be said that there is no major problem even if the combination of the same kind of statistical data is treated as one color, and even if this is converted into Lab data, the information as the statistical value is hardly deteriorated. Also, by converting the combination of statistical data of the same type into Lab data, various statistical data generated for each element color RGB of the input image data can be efficiently converted into information on the Lab color space.

In S250, the computer 20 stores the statistical data 24c after being converted into values in the Lab color space as described above in the HD 24.
The order of the statistical data generation process described in each step of FIG. 6 is not limited to the order of S200 to S230 described above, and may be another order. In addition to the statistical data described above, the computer 20 can generate various statistical values related to the input image data.

(4) Correction amount calculation and correction processing FIG. 10 is a part of image processing executed by the computer 20 according to the APL 25, and mainly shows processing executed according to the correction amount acquisition unit 25d and the correction unit 25e in a flowchart. Yes.
In S300, the computer 20 determines the type of the output color space. The output color space is a color space (working color space) used when the correction amount calculation by the correction amount acquisition unit 25d and the correction processing of the image data by the correction unit 25e are executed. Therefore, the computer 20 determines what the output color space is prior to calculating the correction amount.

  The computer 20 determines the type of the output color space, for example, by accepting designation of the output color space from the outside via a predetermined user interface (UI). Alternatively, when information that predefines the type of the output color space as a default exists in a predetermined storage medium such as the HD 24, the type of the output color space is determined by reading the information. Alternatively, as in S100 above, when the auxiliary information described in the header of the input image data is read and information specifying the output color space exists in the auxiliary information, the output color space is determined based on the information. Determine what it is.

  In S310, the computer 20 determines the type of correction to be applied to the input image data. In the present embodiment, the types of correction that can be executed by the correction unit 25e include memory color correction, human face correction, brightness correction, color balance correction, white balance correction, and the like (details of each correction will be described later). The computer 20 determines the type of correction to be applied to the input image data by accepting designation of the type of correction from the outside via, for example, the UI. In addition, when scene information (landscape image, night scene image, portrait image, etc.) at the time of shooting is described in the header of the input image data, the computer 20 automatically selects the type of correction according to the scene information. It may be determined, or a shooting scene of the input image data may be determined by a predetermined scene determination program, and the type of correction may be automatically determined according to the determination result. Further, the type of correction may be automatically determined according to the type of scene of the input image data designated from the outside via the UI.

In S320, the computer 20 reads the correction target value 24b necessary for realizing the correction according to the type determined in S310 from a predetermined storage medium such as the HD 24.
In S330, the computer 20 converts the read correction target value 24b into data on the output color space determined in S300. In the present embodiment, the correction target value 24b is stored as Lab data regardless of the output color space. Therefore, color space conversion is performed on the correction target value 24b in order to perform correction amount calculation in the output color space.

  FIG. 11 shows an example when the color space of the correction target value 24b is converted to the output color space. In the figure, in S310, the memory color (green) correction, the memory color (sky blue) correction, the memory color (skin color) correction, the memory color (red) correction, the human face correction, and the brightness correction are performed. A state of conversion of the correction target value 24b that is performed when it is determined to be executed is shown. In this case, in S320, among the various correction target values 24b, the target value for memory color (green) correction, the target value for memory color (sky blue) correction, the target value for memory color (skin color) correction, and the memory color ( The target value for red correction, the target value for human face correction, and the target value for brightness correction (brightness target value) are read out.

  This figure is an example of conversion processing when the output color space is set to the wRGB color space. In this case, in S330, the computer 20 converts each Lab data as the read target value into the Lab color space. To RGB data using the above-mentioned ICC profile that defines the conversion correspondence from the wRGB color space to the wRGB color space. Of course, when the output color space is the sRGB color space, each Lab data as the target value is converted into RGB data using the ICC profile that defines the conversion correspondence from the Lab color space to the sRGB color space. Convert.

  In S340, the computer 20 reads from the HD 24 the statistical data 24c necessary for executing the correction related to the type determined in S310. When the description corresponding to the example of FIG. 11 is given, the computer 20 includes the statistical data for the memory color (green) and the statistics for the memory color (sky blue) in the statistical data 24c (Lab data) shown in FIG. Statistical data on data, memory color (skin color), statistical data on memory color (red), statistical data on human face, and statistical data on each area of input image data, average of each area Lab data obtained by converting values Rav, Gav, and Bav are read out.

  In S350, the computer 20 converts the read statistical data 24c into RGB data in the output color space determined in S300. Here again, when the output color space is the wRGB color space, each Lab data as statistical data is converted into RGB data using the above-described ICC profile that defines the conversion correspondence from the Lab color space to the wRGB color space. When the output color space is the sRGB color space, each Lab data as the statistical data is converted into RGB using the ICC profile that defines the conversion correspondence from the Lab color space to the sRGB color space. Convert to data. Note that the computer 20 acquires a predetermined statistical value of input image data and a statistical value expressed by a predetermined output color space as a part of the function in executing the processing from S100 to S350. It can be said that a statistical value acquisition means is realized.

In S360, the correction amount is calculated based on the comparison result between the corrected correction target value and the statistical data.
For example, the target values for memory color (green) correction after conversion in S330 are Rgs, Ggs, and Bgs, and the statistical data for the memory color (green) after conversion in S350 is Rg, Gg, and Bg. In some cases, correction amounts ΔRg, ΔGg, and ΔBg for performing memory color (green) correction on the input image data are obtained as shown in equations (1) to (3).
ΔRg = αg (Rgs−Rg) (1)
ΔGg = αg (Ggs−Gg) (2)
ΔBg = αg (Bgs−Bg) (3)

αg is a weighting coefficient set for the memory color (green) correction, and is a coefficient that defines the degree of application of the memory color (green) correction to the input image data. Such a weighting coefficient is also recorded in a predetermined storage medium such as the HD 24 together with the correction target value 24b.
Correction amounts ΔRb, ΔGb, ΔBb for performing memory color (sky color) correction, correction amounts ΔRs, ΔGs, ΔBs for performing memory color (skin color) correction, and correction amounts for performing memory color (red) correction With respect to ΔRr, ΔGr, ΔBr and correction amounts ΔRf, ΔGf, ΔBf for performing human face correction, like the correction amounts ΔRg, ΔGg, ΔBg, RGB data as correction target values and statistical data It is obtained by multiplying the difference for each element color from the RGB data by a weighting factor α set in advance according to the correction type.

Moreover, the computer 20 calculates | requires the correction amount for brightness correction as follows. As can be seen from FIG. 11, the target value for brightness correction stored in advance as Lab data is converted into RGB data by the process of S330. First, the computer 20 obtains a luminance Y (referred to as a correction reference luminance Ys) from RGB data as a target value for this brightness correction. In general, the luminance Y is calculated using the RGB weighted integration formula (4).
Y = 0.30R + 0.59G + 0.11B (4)
It is possible to calculate by such as. However, in this embodiment, the Lab data as the brightness correction target value is set to a value such that the RGB gradation values are equal when converted to RGB data. For this reason, the computer 20 sets the RGB gradation value (any gradation value of R, G, and B) as the target value for brightness correction after the conversion as the correction reference luminance Ys.

Next, the computer 20 determines the brightness of the input image data used for comparison with the corrected reference luminance Ys based on the statistical data (RGB data) for each region after the conversion in S350. In this case, first, the luminance Yi is calculated from the statistical data (RGB data) for each region. Luminance can be calculated by the above equation (4). Statistical data (RGB data) for each region is obtained by converting average values Rav, Gav, Bav for each RGB of each region of the input image data into Lab data, and converting the Lab data into RGB data depending on the output color space. Therefore, it can be said that each luminance Yi represents a substantially average value of the luminance of each region when the input image data is represented in the output color space. Note that the statistical data for each region corresponds to a statistical value representing the brightness of the input image data.
After obtaining the luminance Yi of each area, the computer 20 then calculates the average luminance value Yav (brightness of the input image data) of the input image data based on all or part of the luminance Yi. Here, as an example, the luminance average value Yav is calculated with emphasis on the luminance Yi in the substantially central region of the image.

  FIG. 12 is a diagram for explaining the calculation process of the luminance average value Yav. This figure shows a state in which the input image data D is divided into a plurality of regions T as in FIG. Further, as described above, the luminance Yi (i = 1 to 25 in the figure) is obtained for each region T here. The computer 20 divides the input image data into a central area and areas other than the central area (referred to as edge areas), calculates the difference between the luminance of the central area and the luminance of each area T of the edge area, and the magnitude of each difference. Are compared with a predetermined threshold value TH. As an example, a range formed by nine regions T in the center of the image among 25 regions is defined as a central region, and 16 regions T outside the central region are defined as edge regions.

  In the case of the luminance of the central region, the average value of the luminance Yi of each region T in the central region may be indicated, or the luminance Yi of a specific region T in the central region may be indicated. Here, one region T in the central region is set as the target block T1, and the difference between the luminance Yn of the target block T1 and the luminance (Y1, Y2, Y3...) Of each region T of the edge region is calculated. Then, among the regions T of the edge region, those where the magnitude of the calculated difference exceeds the threshold value TH is excluded from the calculation target of the luminance average value Yav, and each region T of the edge region that is not excluded. An average value is calculated on the basis of the luminance Yi and Yi of each region T in the central region, and this calculation result is defined as a luminance average value Yav.

  However, the method of calculating the luminance average value Yav with emphasis on the luminance of the substantially central region of the image is not limited to the above method. For example, the luminance Yi of each region T is integrated by giving a weighting coefficient (a value of 0 to 1) according to the position in the image of each region T, and the integration result is the average luminance value of the input image data. Yav may be used. It is assumed that this weighting factor gives a larger value to the region T closer to the center of the image. In addition, the sum of the weighting coefficients given to each region T is set to 1.

After calculating the correction reference luminance Ys and the average luminance value Yav as described above, the computer 20 obtains a brightness correction amount ΔY for performing brightness correction on the input image data as shown in Expression (5).
ΔY = αy (Ys−Yav) (5)
αy is a weighting coefficient set for brightness correction, and the weighting coefficient is also recorded in a predetermined storage medium such as the HD 24 together with the correction target value 24b.
Through the above processing, correction amount calculation (memory color (green) correction, memory color (sky blue) correction, memory color (skin color) correction, memory color (red) correction, human face correction, brightness, and brightness) in S360. The calculation of each correction amount when performing correction) is completed.

  Here, the reason why the luminance of the substantially central region of the image is emphasized when obtaining the luminance average value Yav of the input image data is as follows. In the approximate center region of the image, for example, a particularly important image is often represented, for example, a human face. Accordingly, the luminance of the input image data (luminance average value Yav) is calculated by placing importance on the luminance of the substantially central region where an important subject is likely to be represented rather than the luminance of the edge region. An optimum brightness correction amount ΔY for the image can be obtained. Also, when the image is dark in the center part and bright around the center part, such as an image taken under backlight conditions, the brightness of the dark center part is set by setting a correction amount ΔY that emphasizes the brightness of the center part. It is possible to perform optimum brightness correction that increases the brightness.

  In S370, the computer 20 generates a correction LUT having a correction degree corresponding to the correction amount. For example, a tone curve is generated as a correction LUT. If the correction amount for each RGB is calculated in S360, a tone curve for each RGB is generated. In the above-described example, the correction amounts ΔRg, ΔGg, ΔBg for the memory color (green), the correction amounts ΔRb, ΔGb, ΔBb for the memory color (sky blue), the correction amount ΔRs for the memory color (skin color), ΔGs, ΔBs, memory color (red) correction amounts ΔRr, ΔGr, ΔBr, and correction amounts ΔRf, ΔGf, ΔBf for human face correction were calculated. Therefore, the computer 20 integrates the respective correction amounts for each RGB, and sets the integrated values as total correction amounts ΔR, ΔG, ΔB for each RGB. That is, in this case, ΔR = ΔRg + ΔRb + ΔRs + ΔRr + ΔRf, ΔG = ΔGg + ΔGb + ΔGs + ΔGr + ΔGf, and ΔB = ΔBg + ΔBb + ΔBs + ΔBr + ΔBf. Then, tone curves corresponding to the correction amounts ΔR, ΔG, and ΔB are generated.

FIG. 13 shows an example of a tone curve. The tone curve C shown in the figure is an example of an LUT that corrects the input tone value (0 to 255) to the output tone value (0 to 255) for the element color R, and the degree of the curve is adjusted as described above. It is generated by determining according to the amount ΔR. That is, correction is performed by adding the correction amount ΔR to a predetermined control point CP on the straight line graph F where the input gradation value = output gradation value shown in the figure, and the corrected control point CP ′ and the straight line graph F are corrected. A curve including both ends of the curve is calculated using spline interpolation, and the calculated curve is set as a tone curve C. The position of the control point CP is not particularly limited. For example, when the correction amount is a positive value, it corresponds to one input gradation value lower than the intermediate value (128) of the input gradation range. If the position on the straight line graph F is the control point CP and the correction amount is a negative value, the position on the straight line graph F corresponding to one input gradation value on the higher gradation side than the intermediate value is set. It can be a control point CP. Of course, the control point CP may be determined based on the RGB gradation values of the statistical data after the conversion in S350. In this way, the computer 20 also generates tone curves for the element colors G and B, respectively.
If the correction amount ΔY for brightness correction has already been calculated as described above, a tone curve corresponding to this ΔY is also generated and used as the brightness correction LUT.

  In S380, the computer 20 converts the color space of the input image data into an output color space as necessary. Such conversion processing is performed only when the input color space and the output color space are different, and when the input color space and the output color space are the same, the processing of S380 is skipped. The computer 20 has already determined the type of the input color space and the type of the output color space in S100 and S300. Therefore, referring to the determination result, for example, when the input color space is the sRGB color space and the output color space is the wRGB color space, the computer 20 sets a conversion profile for converting the sRGB color space into the wRGB color space. The data is read from a predetermined storage medium such as the HD 24, and the RGB data of each pixel of the input image data is converted into RGB data in the output color space using the read conversion profile. Conversely, when the input color space is the wRGB color space and the output color space is the sRGB color space, the computer 20 reads a conversion profile for converting the wRGB color space into the sRGB color space from a predetermined storage medium such as HD24. The RGB data of each pixel of the input image data is converted into RGB data in the output color space using the read conversion profile.

In S390, the computer 20 corrects the input image data expressed in the output color space by inputting the RGB gradation values of each pixel into the correction LUT for each RGB generated in S370. If the brightness correction LUT has already been generated, the brightness Y of each pixel is corrected by being input to the brightness correction LUT.
In S400, the computer 20 outputs the image data after performing all the corrections related to the correction type determined in S310, and ends the process of FIG.

Here, among the various statistical data shown in FIG. 9, statistical data for low saturation colors, statistical data representing the maximum value of the histogram, and the like are also used for correction of the input image data as follows. it can.
The statistical data (Lab data) for the low saturation color is obtained by converting an average value for each RGB calculated based on a pixel group belonging to a predetermined color gamut having a low saturation such as black or white. Therefore, when the statistical data (Lab data) for the low saturation color is converted into RGB data in the output color space, if there is a shift between the RGB, the color balance of the input image data is shifted. It can be said. Therefore, if the computer 20 decides to perform color balance correction in S310, the computer 20 converts the statistical data (Lab data) for the low chroma color into RGB data in the output color space, and after the conversion. The amount of deviation between RGB in RGB data (for example, the difference of R with respect to G and the difference of B with respect to G) is calculated, and RGB of each pixel of the input image data is corrected so as to cancel the calculated difference. Thus, the color balance of the input image data may be adjusted.

Further, when the statistical data (Lab data) representing the maximum value of the histogram is converted into RGB data in the output color space, if there is a shift between the RGB, the shift of each RGB maximum value of the input image data In other words, it can be said that there is a deviation in white balance. Therefore, if the computer 20 decides to perform white balance correction in S310, the computer 20 converts the statistical data (Lab data) representing the maximum value of the histogram into RGB data in the output color space, and the conversion A shift between RGB in the later RGB data (for example, the difference of each RGB with respect to a predetermined reference gradation value set as a maximum value common to RGB) is obtained. Then, based on such a difference, the white balance of the input image data may be adjusted by performing correction so that the RGB maximum values of the input image data substantially match.
Note that the color balance correction and the white balance correction are corrections for correcting a deviation between RGB, and therefore there is no target value for such correction.

(5) Modification Here, the content of the color gamut definition information 24a may be generated based on the correction target value 24b. As described above, the computer 20 has the target value of each memory color in the form of Lab data as a kind of the correction target value 24b. Therefore, the computer 20 defines a solid having a predetermined size as shown in FIG. 3 for each memory color with the coordinate position in the Lab space of the target value of each memory color as the center, and the solid is defined as lightness L, saturation C, Information defined by each range of hue h may be stored in the HD 24 as color gamut definition information 24a for each storage color. If the color gamut definition information 24a is generated based on the correction target value 24b as described above, a pixel that is a generation source of statistical data for memory color correction can be accurately extracted from the pixels of the input image data. .

Various methods of using statistical data stored as information in the device-independent color space can be considered other than the above-described examples.
For example, the computer 20 acquires a plurality of input image data (for example, landscape image data) having a common image scene, generates statistical data for each input image data, and generates from each input image data. The statistical data may be averaged for each element color and each type, and the averaged statistical data may be stored in a predetermined storage medium such as the HD 24 as standard statistical data for a landscape image. The standard statistical data for the landscape image is general statistical data for correcting the landscape image. Of course, the computer 20 can obtain standard statistical data for various scenes such as a portrait and a night view in addition to the landscape.

  Then, after acquiring the standard statistical data for each scene, the computer 20 determines the scene of the input image data when the input image data to be corrected is input, and the standard corresponding to the corresponding scene. Read statistical data. As described above, the scene determination of the input image data can be performed by information described in the header of the input image data, a predetermined scene determination program, or scene specification by the user via the UI. . Then, a correction amount is calculated according to the comparison result between the read standard statistical data and the correction target value 24b, and correction processing according to the calculated correction amount is performed. As a result, the computer 20 can perform optimal correction according to the type of the scene without performing extraction of memory color pixels or generation of statistical data with respect to newly input image data to be corrected. .

  Further, the computer 20 may provide standard statistical data for each scene to each image output device such as the printer 31 and the display 32 via various portable storage media and communication lines. As a result, the printer 31 and the display 32 also use the standard statistical data for each scene to calculate an appropriate correction amount according to each scene, and perform correction according to the correction amount. It is possible to perform printing and video display based on subsequent image data. Even in such a case, since the statistical data is provided as information on the device-independent color space, it can be easily handled by performing color space conversion as necessary on the side of each image output device, and is highly versatile. .

(6) Summary As described above, according to the present invention, the color gamut definition information 24a in which the memory color range is defined in the device-independent color space is held, and the memory color pixel is extracted from the input image data. The color gamut defined by the color gamut definition information 24a is converted into a color gamut that depends on the color space of the input image data. That is, even in a situation where the color space of the input image data is different such as the sRGB color space or the wRGB color space, the color gamut definition information 24a defined for a certain memory color is converted into a color gamut that matches the color space of the input image data. To do. Therefore, it is possible to always accurately extract pixels corresponding to the memory color regardless of the color space of the input image data. Further, it is not necessary to prepare color gamut definition information in advance for each color space that can be adopted by input image data as in the prior art.

The color gamut conversion processing according to the present invention only converts a predetermined number of coordinates on the outline of the color gamut defined by the color gamut definition information 24a into information depending on the color space of the input image data. Compared to the case where all the pixels are converted to the color gamut definition information color system, the amount of calculation processing is remarkably reduced, and the time required for pixel extraction can be greatly shortened. In particular, since only one coordinate point is converted into information depending on the color space of the input image data for each surface constituting the color gamut defined by the color gamut definition information 24a, the amount of calculation processing is very small.
Also, when expressing a color gamut defined in a certain color space by converting it to another color space, the shape of the color gamut after conversion may be greatly distorted, but in this embodiment, the conversion to be converted The reference point P is set as the center position for each surface constituting the color gamut of the color gamut definition information 24a. Therefore, it is possible to prevent a large distortion from occurring in the color gamut defined by the coordinate points after conversion, and as a result, it is possible to realize more accurate extraction of pixels related to the memory color.

  Further, in the present invention, when calculating various statistical data from the input image data, the calculation is performed while maintaining the color space, and the calculated statistical data is converted into information on the device-independent color space and stored in a predetermined storage medium. I tried to do it. Therefore, the stored statistical data is used after color space conversion if necessary even if the work color space (output color space) used for calculating the correction amount is different as in the sRGB color space or the wRGB color space. It can be said that the information is highly versatile. In addition, the acquisition of statistical data eliminates the need to convert the input image data itself into work color space information as in the prior art, and the required color space conversion work requires a limited amount of information. It just converts the data into device-independent color space information. Therefore, the calculation processing amount and time required to obtain statistical data are greatly reduced as compared with the conventional case. In addition, since it is not necessary to determine the type of work color space (output color space) in the process of acquiring statistical data, the process of acquiring statistical data can be accelerated.

  Further, when the input image data is converted into information on the working color space, even if the same input image data is converted according to the same conversion rule (conversion profile), the platform on which the APL 25 that executes the conversion process operates, etc. Depending on the difference, the conversion result may be different. If the conversion results are different as described above, the contents of the statistical data obtained based on the conversion results are naturally different. Therefore, it is desirable to obtain statistical data while maintaining the color space of the input image data as much as possible. According to the present invention, when calculating the statistical data from the input image data, the calculation is performed while maintaining the color space. Therefore, the input image data is converted into the working color space before calculating the statistical data. The error of the statistical data that may have occurred due to is eliminated.

  In the present invention, the correction target value 24b to be compared with the statistical data is defined in the device-independent color space, and the correction target value 24b is converted into information on the output color space when calculating the correction amount. . Accordingly, even when the output color space is different such as the sRGB color space or the wRGB color space, it is possible to calculate the correct correction amount by converting the correction target value 24b into information on the output color space. . Further, unlike the prior art, the trouble of preparing the correction target value 24b in advance for each output color space that can be adopted is eliminated, and memory resources are saved.

1 is a block diagram showing an image processing apparatus according to an embodiment of the present invention. The flowchart which showed the content of the pixel extraction process. The figure which showed the color gamut which color gamut definition information defines in Lab color space. The figure which showed a mode that the conversion reference point was set to the color gamut which color gamut definition information defines. The figure which showed the color gamut after conversion in HSV color space. The flowchart which showed the content of the acquisition process of statistical data. The figure which showed an example of the histogram for every RGB. The figure which showed an example of a mode that input image data was divided | segmented. The figure which showed an example at the time of converting statistical data into Lab data. The flowchart which showed the content of calculation of correction amount, and a correction process. The figure which showed an example at the time of converting a correction target value into the data on output color space. The figure which showed the brightness | luminance of each area | region which divided | segmented input image data. The figure which showed an example of the tone curve.

Explanation of symbols

DESCRIPTION OF SYMBOLS 20 ... Computer, 21 ... CPU, 22 ... ROM, 23 ... RAM, 24 ... Hard disk, 24a ... Color gamut definition information, 24b ... Correction target value, 24c ... Statistical data, 25 ... APL, 25a ... Image data acquisition part, 25b ... pixel extraction section, 25c ... statistical data acquisition section, 25c1 ... statistical data generation section, 25c2, 25d2 ... statistical data conversion section, 25d ... correction amount acquisition section, 25d1 ... target value conversion section, 25d3 ... correction amount determination section, 25e ... correction unit

Claims (8)

Statistical value generation means for acquiring input image data expressed by a device-dependent color space and generating predetermined statistical values from the input image data while maintaining the color space;
An image processing apparatus comprising: a statistical value conversion unit configured to convert the generated statistical value into color information expressed by a device-independent color space. Converts a specific color gamut defined in the device-independent color space to a color gamut that depends on the color space of the input image data, and extracts pixels belonging to the converted color gamut from the pixels that make up the input image data A pixel extraction unit that
The image processing apparatus according to claim 1, wherein the statistical value generation unit generates a statistical value based on the extracted pixel group.   3. The image processing apparatus according to claim 1, wherein the statistical value generation unit divides the input image data into a plurality of regions and generates a statistical value for each of the divided regions. .   The statistical value generation means detects a human face existing in the input image data, and generates a statistical value based on a pixel group in the detected human face region. Item 4. The image processing device according to any one of Items 3 to 4.   The statistical value generation means generates a statistical value for each element color constituting the input image data, and the statistical value conversion means regards a combination of the same kind of statistical values for each element color as one color, and uses the combination as a device. The image processing apparatus according to claim 1, wherein the image processing apparatus converts color information expressed by an independent color space.   The statistical value generation means acquires a plurality of input image data having a common image type, generates a statistical value from each of the plurality of input image data, averages the statistical value for each of the plurality of input image data, The statistical value conversion means converts the averaged statistical value into color information expressed by a device-independent color space, and stores the converted color information as a statistical value for the common image type as a predetermined value. 6. The image processing apparatus according to claim 1, wherein the image processing apparatus is stored in an area. Obtaining the input image data represented by the device-dependent color space, and generating a predetermined statistical value from the input image data while maintaining the color space; and
An image processing method comprising: a statistical value conversion step of converting the generated statistical value into color information expressed by a device-independent color space. A statistical value generation function for acquiring input image data expressed by a device-dependent color space and generating predetermined statistical values from the input image data in a state in which the color space is maintained;
An image processing program causing a computer to execute a statistical value conversion function for converting the generated statistical value into color information expressed by a device-independent color space.

Images