JP4505356B2 - Image processing apparatus, image processing method and program thereof - Google Patents

Description

  The present invention relates to an image processing apparatus, an image processing method, and a program for compressing the dynamic range of a color image.

  Conventionally, an image obtained by photoelectrically reading an image obtained by photographing a subject with an imaging device such as a digital camera, an image recorded on a photographic film such as a negative film or a reversal film, or a printed matter. Is reproduced on a reproducing apparatus such as a printer. In this way, when the image is reproduced on the reproduction device, in order to prevent the bright portion and / or the dark portion on the image from being crushed, the image is set so that the dynamic range of the image is within the dynamic range of the reproduction device. The dynamic range of compression has been done.

  In such a dynamic range compression method, a filtering process using a low-pass filter is performed on an image to create a blurred image representing only a structure having a low spatial frequency in the image, and the dynamic range is compressed using the blurred image. Thus, there has been proposed a method for suppressing the collapse of both the bright and dark portions of the image while keeping the fine texture contrast in the bright and dark portions included in the image (see Patent Documents 1 and 2). Specifically, when compressing the low density area of the image, a compression table for compressing the low density area of the image is obtained based on the dynamic range of the image as shown in FIG. The gray value of each pixel is calculated from the RGB color information. In FIG. 8, Y0 is a reference pixel value representing a boundary for compressing the dynamic range of the input pixel value. The horizontal axis indicates that the concentration is lower as it goes to the left. Then, referring to the compression table, the correction value (referred to as C30) of the color information of each pixel is determined from the gray value of each pixel, and as shown in the following equation (1), the color information (that is, each pixel) By adding the correction value C30 to the RGB color information R30, G30, B30), a processed image in which the color information of each pixel is R31, G31, B31 is obtained.

R31 = R30 + C30
G31 = G30 + C30 (1)
B31 = B30 + C30
JP-A-9-130609 JP 2001-245153 A

  However, in the methods described in Patent Documents 1 and 2, since the correction value is determined based on the gray value of each pixel, if the dynamic range of the image is compressed by the determined correction value, the correction value is included in the image. There is a problem that the high-saturation color that is generated becomes saturated and collapses. Hereinafter, this problem will be described.

  FIG. 9 is a diagram for explaining a conventional problem. In FIG. 9, a solid line indicates a gray gradation of an input image in which a bright portion having a low density in the image is outside the dynamic range (density reproduction area) of the output device by AE processing. The input image having such a gradation has a high output density on the lightness side (ie, it is darkened) as indicated by a broken line with reference to the reference pixel value Y0, and each input image has a dynamic range within the output device. A compression table for correcting the color information of the pixel is obtained, a correction value is obtained by referring to the compression table from the gray value of each pixel of the input image, and the correction value is added to all the color information of each pixel. A processed image is obtained.

  Here, for simplicity, it is assumed that the input image is composed of two colors R and B, and the gray value at each pixel is calculated as an average value of the R signal and the B signal. As shown in FIG. 9, when the R signal at a certain pixel P30 is R30 and the B signal is B30, the gray value Y30 of the pixel P30 is calculated by (R30 + B30) / 2. At this time, the correction value of the pixel P30 is determined as C30. Therefore, by adding the correction value C30 to the R signal R30 and the B signal B30 of the pixel P30, the R signal and the B signal of the pixel P30 are corrected to R31 and B31, respectively.

  As shown in FIG. 9, the modified B signal B31 is within the dynamic range of the output device, but the modified R signal R31 is not within the dynamic range of the output device despite being modified. For this reason, when the processed image is reproduced in the output device, the R signal R31 is clipped to the brightest value that can be reproduced by the output device. For example, when the dynamic range of the output device is 8 bits (0 to 255), if R31 and B31 are pixel values of 300 and 200, respectively, R31 is clipped to 255. For this reason, the saturation of the pixel P30 is saturated by the dynamic range compression process, and as a result, the saturation is reduced in the reproduced image.

  The present invention has been made in view of the above circumstances, and an object thereof is to compress the dynamic range of an image so that the saturation does not decrease.

An image processing apparatus according to the present invention includes an image input unit that inputs a color image including pixels having a plurality of color information,
A corrected image creating means for creating a corrected image by correcting the density of color information of pixels of the color image so as to increase the brightness of the color image;
A calculation method for calculating a correction value for correcting a density value in a low density region among density values of color information of pixels of the corrected image so that a density value exceeding a predetermined dynamic range is within the dynamic range. A plurality of different correction value calculation means,
When the density value of the color information of the pixel of the corrected image is within the low density range, the correction value is obtained from the plurality of correction value calculating means at a ratio according to the color of the pixel of the color image corresponding to the pixel of the corrected image. And dynamic range compression means for performing dynamic range compression processing using a plurality of correction values obtained.

The image processing method of the present invention is
An image input step of inputting a color image composed of pixels having a plurality of color information;
A corrected image creating step of creating a corrected image in which the density of color information of pixels of the color image is corrected so as to increase the brightness of the color image;
A calculation method for calculating a correction value for correcting a density value in a low density region among density values of color information of pixels of the corrected image so that a density value exceeding a predetermined dynamic range is within the dynamic range. A plurality of different correction value calculation steps,
When the density value of the color information of the pixel of the corrected image is within the low density range, the correction value is obtained from the plurality of correction value calculating steps at a ratio according to the color of the pixel of the color image corresponding to the pixel of the corrected image. And a dynamic range compression step for performing a dynamic range compression process using the plurality of correction values obtained.

Furthermore, the program of the present invention is a computer,
Image input means for inputting a color image composed of pixels having a plurality of color information;
A corrected image creating means for creating a corrected image by correcting the density of color information of pixels of the color image so as to increase the brightness of the color image;
A calculation method for calculating a correction value for correcting a density value in a low density region among density values of color information of pixels of the corrected image so that a density value exceeding a predetermined dynamic range is within the dynamic range. A plurality of different correction value calculation means,
When the density value of the color information of the pixel of the corrected image is within the low density range, the correction value is obtained from the plurality of correction value calculating means at a ratio according to the color of the pixel of the color image corresponding to the pixel of the corrected image. It is characterized by functioning as dynamic range compression means for performing dynamic range compression processing using a plurality of correction values.

  Here, the color information of each pixel is “ITU Rec. In some cases, it represents the luminance (taken a value of 0 to 1) defined by 709 or the like, and in some cases represents the density obtained by log conversion, but in the case where the color signal represents the luminance when performing “dynamic range compression processing” Is multiplied by the correction value, and if the color signal represents density, the correction value may be added. Therefore, “the dynamic range compression process is performed using the correction value” means that the correction value is multiplied when the color information represents luminance, and the correction value is added when the color signal represents density. To do.

  The “predetermined dynamic range” represents the reproduction ability of a predetermined image reproduction device such as a printer or a monitor to reproduce a signal.

  “Correct the density value in the low density range exceeding the predetermined dynamic range so that it falls within the dynamic range” means that the color image has a plurality of color information of each pixel not exceeding the dynamic range. If one of them exceeds the dynamic range, the color information is corrected to be within the dynamic range. “Correction to be within the dynamic range” means to correct each color information value so that it is within the dynamic range, but most of the density values that exceed the dynamic range are dynamic. As long as it falls within the range, a value exceeding the dynamic range may remain in a part (part where the density value is extremely low).

If one of the plurality of correction value calculation means calculates a correction value for skin color, and one of the plurality of correction value calculation means calculates a correction value for non-skin color,
The dynamic range compression means determines the skin color correction value and the non-skin color according to the skin color rate representing the skin color likelihood of the color of the color image pixel corresponding to the pixel of the correction image and the non-skin color rate representing the non skin color. The dynamic range compression processing may be performed using a correction value obtained by weighted averaging the correction values for skin color.

  According to the present invention, a correction value for correcting a value in a low density region exceeding the dynamic range so as to be within the dynamic range is obtained by a plurality of calculation methods, and a plurality of corrections are performed at a rate according to the color of the pixel of the color image. If dynamic range compression processing is performed using values, it is possible to perform dynamic range compression suitable for the color of the target object.

  Further, by changing the correction value depending on the skin color and the non-skin color, it is possible to perform dynamic range compression on an image in which a person's face is photographed so that the person's face is finished in a natural feeling.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a schematic block diagram showing the configuration of an image processing apparatus according to an embodiment of the present invention. As shown in FIG. 1, the image processing apparatus 1 according to the present embodiment includes an image input unit 2 that inputs a color image (hereinafter referred to as an original image) S0 composed of RGB color data, and the brightness of the entire image of the original image S0. The corrected image generating means 3 for generating the corrected image S1 by performing AE correction so as to increase the number of correction values, the correction value calculating means 4 for calculating the correction values for dynamic len compression, and the pixels of the corrected image. Dynamic range compression means 5 that generates processed image data S2 that has been subjected to dynamic range compression processing by adding correction values obtained from a plurality of correction value calculation means at a rate according to the color of the pixel of the original image. . The image processing apparatus 1 is connected to an image output unit 7 such as a printer or a monitor that reproduces the processed image data S2 subjected to the dynamic range compression process. In the present embodiment, the image output unit 7 is a printer that obtains the print P of the processed image data S2.

  The image input means 2 is a media drive that reads the original image data S0 from the medium, and various interfaces that accept input of the image data S0 transmitted via the network. The original image data S0 may be acquired by an imaging device such as a digital camera, or may be acquired by photoelectrically reading an image recorded on a film or a document.

The corrected image creation means 3 performs AE correction on the original image data S0 to create corrected image data S1. For example, when the subject included in the original image S0 is dark due to backlight or the like, the density of the entire original image S0 is lowered to brighten the entire original image S0. Specifically, the corrected image S1 is created by performing AE correction processing by adding a fixed value to each RGB color information in each pixel of the original image S0. When representing the uncorrected output density of the relationship between the input value and the image output section 7 from the corrected image data in the graph consists of l ₀ before AE correction as shown in FIG. 2 to l ₁ after AE correction.

  Further, the image processing apparatus 1 is input with information on the minimum density Dmin that can be reproduced by the image output unit 7, and whether or not the corrected image S1 includes a pixel having color information larger than the minimum density Dmin. If the determination is negative and the determination is negative, the corrected image data S1 is output to the image output unit 7 as it is. If the determination is positive, the correction value calculation means 4 calculates the correction value.

The correction value calculation means 4 calculates the correction value as follows. First, the gray value Y of each pixel of the corrected image S1 is calculated by the following equation (2). R1, G1, and B1 are RGB color information of each pixel of the corrected image S1.
Y = 0.3125R1 + 0.3750G1 + 0.3125B1 (2)
Next, the correction value calculation unit 4 obtains a gray value Y histogram. FIG. 3 is a diagram showing a histogram H of gray values Y. In the histogram H, the correction value calculation unit 4 obtains the density at which the cumulative relative frequency from the highlight side is 3% as the highlight gray value Yh of the image data S0. When the corrected image S1 is represented by 8 bits, each color information of RGB takes a value of 0 to 255, but the pixel value of the bright portion may be larger than 255 by the AE correction process.

On the other hand, information on the minimum density Dmin that can be reproduced by the image output unit 7 is input to the correction value calculation means 4, and the highlight gray value Yh is compared with the highlight gray value Yh. When the highlight gray value Yh becomes brighter than the minimum density Dmin, the highlight gray value Yh is set within the minimum density Dmin around the preset reference gray value Y0. For calculating the dynamic range compression ratio. That is, in the graph showing the relationship between the input value and the output density shown in FIG. 4A, when the low density region in the gray gradation of the corrected image S1 becomes brighter than the minimum density Dmin as shown by the solid line, FIG. The dynamic range compression rate P0 is calculated so that the low density region falls within the minimum density Dmin as indicated by the broken line 4 (a). Specifically, the dynamic range compression rate P0 is calculated as shown in the following formula (3).
P0 = (Y0−Yh) / (Y0−Dmin) (3)

  Here, as the value of the reference gray value Y0, for example, when the subject included in the image S0 is a person, a value between 0.50 and 0.70 that is substantially the same density as the skin color (as a ratio to the number of bits). ), Preferably 0.6.

  Further, the correction value calculation means 4 generates a compression table T0 based on the dynamic range compression rate P0. FIG. 4B is a diagram illustrating an example of a compression table. As shown in FIG. 4, in the compression table T0, Y0 is the reference gray value described above, and the slope of the straight line portion is 1 / P0-1. By using such a compression table, when the low density region in the gray gradation of the image S0 is smaller than the minimum density Dmin as shown by the solid line in FIG. 4A, it is shown by the broken line in FIG. As described above, the low concentration region falls within the minimum concentration Dmin.

  In the compression table T0, when the slope changes (that is, the point of the reference gray value Y0) changes discontinuously, an artifact may occur in the processed image. Therefore, a function is set so that the slope is continuous. As a result, the occurrence of artifacts can be prevented.

  The correction value calculation means 4 calculates the correction value C0 of the color information of each pixel of the corrected image data S1 as follows.

  First, the first calculation method will be described. In the first calculation method, the correction value is obtained from the color information of the brightest color among the RGB color information in each pixel of the image S0. Here, the brightest color is a color having the largest value among the RGB color information in each pixel of the image S0. For example, when the image data S0 is 8 bits, each color information of RGB takes a value of 0 to 255, but the pixel value of the bright part may be larger than 255 by the AE process, so The color information may have a value larger than 255.

  For example, when the brightest color is R among the RGB color information R1, G1, B1 in a certain pixel P1, the correction value corresponding to the color information R1 on the horizontal axis in the compression table T0 is set to all the color information R1 in the pixel P1. , G1, B1 is determined as a correction value C0.

  When the color information of the brightest color is equal to or less than the reference gray value Y0, the correction value C0 is 0. For this reason, it is determined whether or not the color information of the brightest color of the pixel to be processed is larger than the reference gray value Y0. Only when this determination is affirmative, the correction value C0 is referred to the compression table T0. decide.

When this correction value C0 is used, processed color information R2, G2, B2 can be obtained by adding to all the color information R1, G1, B1 of each pixel.
R2 = R1 + C0 (4)
G2 = G1 + C0 (5)
B2 = B1 + C0 (6)

  Here, for simplicity, it is assumed that the original image S0 and the corrected image S1 are composed of two colors R and B, and a description will be given using a graph showing the relationship between the input value and the output density shown in FIG. If the value of R in the pixel P1 is R1 and B is B1, the brightest color is R. Therefore, the correction value of the pixel P1 is determined as C0 which is the correction value of the R value R1. Therefore, by adding the correction value C0 to the other color information values R1 and B1 of the pixel P1, the color information of R and B of the pixel P1 is corrected to R2 and B2, respectively. As a result, both of the corrected R2 and B2 are within the dynamic range of the image output unit 7.

  Such correction is performed in the low density region, and as can be seen from FIG. 5, in the low density region (broken line portion), the change due to gradation becomes gradual and the density of the image is reduced. The skin-colored area where the subject's face is photographed often falls within this low-density area. is there.

  Therefore, the correction value for the skin color region is determined by a second calculation method different from that described above. First, in the second calculation method, the dynamic range compression rate P0 as shown in the equation (3) is calculated as in the first calculation method. In the second calculation method, a correction value different for each color information is obtained without correcting all the color information with the same value.

First, a broken line (dashed line) is drawn between the broken line and the solid line in FIG. 5 to create a graph as shown in FIG. As a first step, a correction value C1 for correcting the color information R and a correction value C2 for correcting the color information B are determined. Correction value C1 is a correction value for correcting color information of R after AE corrected R1 (the value on l ₁₎ to a concentration value of the one-dot chain line l _2, correction value C2, the color of B after AE correction information B1 (the value of the l ₁₎ is a correction value for correcting up to a concentration value of the one-dot chain line l _2. Further, as the second stage, the correction value C3 is determined so that the gray value D (here, for simplicity, the average value of the color information R and B is the gray value) is corrected to the density value on the alternate long and short dash line l _2. Then, both R1 and B1 are further corrected using the correction value C3.

By using the first-stage correction values C1 and C2 and the second-stage correction value C3, all the color information R1 and B1 of each pixel are added to obtain processed color information R2 and B2.
R2 = R1 + C1 + C3 (7)
B2 = B1 + C2 + C3 (8)

  When all the color information is corrected using the same correction value C0 as in the first calculation method, the hue (hue) is maintained. On the other hand, when the correction is performed as in the second calculation method, there is a possibility that the original hue may be slightly different due to the difference between the correction values C1 and C2, but the density due to the difference in gradation is higher than that in the first calculation method. The change of becomes large and the light and shade appears. However, if the difference between the correction values C1 and C2 is made very large and the ratio of the correction value C3 is made small, there is a high possibility that the hue will be considerably different from the original color. Correction values C1, C2, and C3 are calculated so as not to be satisfied.

  The skin color has the largest value in the R color information, but the correction value C3 is calculated on the basis of the gray value (average value of the color information), so that R is a part of the low density region (very low and bright). In some cases, the value may exceed the minimum density Dmin, and the saturation will decrease there. However, even if the saturation decreases in this way, it is a part of the face and does not appear in a wide area. preferable.

  On the other hand, when the second calculation means is used for the sky part, when the sky is very light and blue is close to white, the saturation of the entire sky is lowered and the sky becomes grayish, resulting in an unnatural image. It becomes. Therefore, if the first calculation method is used for such a blue portion, the density value falls within the dynamic range and the saturation does not decrease.

  In other words, the first calculation means is determined according to the color of the color information of the pixel of the original image S0, giving priority to the fact that the hue does not change rather than the density change due to the gradation, while the second The calculation means determines the correction value of each color information so as to be close to the change in density due to the gradation of the original image S0 because the hue does not change. By changing the correction value obtained from each calculation unit according to the target to be corrected, it is possible to correct to a preferable image.

  The dynamic range compression unit 5 determines the correction value obtained by the first calculation unit depending on whether the color of the pixel of the original image S0 corresponding to the pixel of the corrected image S1 is a skin color or a color other than the skin color (non-skin color). And a processed image S2 subjected to dynamic range compression processing using a correction value obtained by weighted averaging of the correction values obtained by the second calculation means.

  Since it is difficult to clearly determine whether or not the color appearing in the image is a skin color, the skin color ratio r representing the skin color is determined in a certain color space. As shown in FIG. 7, the axis (GB) indicating the value obtained by subtracting the pixel value of the color information B from the pixel value of the color information G, and the pixel value of the color information G from the pixel value of the color information R are subtracted. The skin color rate is expressed by the density value on the skin color rate map Map composed of the axis indicating the value (RG), and the skin color rate is r = 1 for the pure white part, and the skin color rate is r = 0 for the black part. Gray represents an intermediate skin color rate (0 <r <1). Further, the non-skin rate is 1-r.

From the color information of the pixel of the original image S0 corresponding to the pixel on the corrected image S1, the values of GB and RG are obtained, and the skin color rate r is obtained from the gray value on the skin color rate map Map. The correction value is obtained by adding the correction value obtained by the first calculation method and the correction value obtained by the second calculation method according to the skin color rate r and the non-skin color rate 1-r. For example, when the pixel of the original image S0 is at the point indicated by the arrow in the skin color rate map Map in FIG. 7 and the skin color rate indicated by the gray value at this point is 0.3, the non-skin rate is 0.7. The correction values of the first calculation method and the second calculation method are weighted and averaged according to the non-skin rate and the non-skin rate to obtain the R correction value D1 and the B correction value D2.
D1 = r (C1 + C3) + (1-r) C0
= 0.3 (C1 + C3) + 0.7 x C0 (9)
D2 = r (C2 + C3) + (1-r) C0
= 0.3 (C2 + C3) + 0.7 × C0 (10)

By using the correction values D1 and D2 and adding to all the color information R1 and B1 of each pixel of the corrected image, processed color information R2 and B2 can be obtained.
R2 = R1 + D1 (11)
B2 = B1 + D2 (12)

  In this way, by calculating the skin color rate and changing the correction value, dynamic range compression that prioritizes gradation is applied to areas where gradation is better even if there is a slight decrease in saturation, such as the face. In a place where it is better not to lower the saturation as in the sky, it is possible to perform dynamic range compression so that the hue does not change.

In the above description, the compression table is calculated from the gray value. However, the compression table may be calculated from the color information of the brightest color. Specifically, the pixel value of the brightest color is extracted from each color information of RGB in each pixel of the correction image S1, and the histogram H1 of the brightest color is obtained from all the pixels as shown in FIG. The calculation unit 4 may obtain a value Ch _3% indicating that the cumulative relative frequency from the highlight side in the histogram H1 is 3% as Yh of the image data S0 shown in FIG.

Further, as shown in FIG. 11A, by calculating the dynamic range compression ratio so that the highlight side maximum value Chmax in the histogram H1 falls within the minimum density Dmin, the color information of all images is obtained for non-skins. The compression table (FIG. 11 (b)) can be determined so as to be within the minimum density Dmin. Alternatively, the maximum value Chmax in FIG. 11A may be a value Ch _3% indicating that the cumulative relative frequency is 3%.

  Moreover, it can operate as an image processing apparatus by installing in a computer using the medium which recorded the program which functions each said means on a computer. This program can also be provided via a network.

As described above in detail, it is possible to prevent the face from becoming unnatural and uncomfortable while keeping the saturation as low as possible.

1 is a schematic block diagram showing the configuration of an image processing apparatus according to an embodiment of the present invention. The figure for demonstrating AE correction processing Diagram showing gray value histogram The figure for demonstrating the production | generation of the compression table which compresses a dynamic range (the 1) Diagram for explaining compression of dynamic range (part 1) Diagram for explaining compression of dynamic range (part 2) Diagram showing skin rate map Diagram showing compression table The figure for demonstrating the conventional dynamic range compression process The figure which shows the histogram of the value of the brightest color Diagram for explaining generation of compression table for compressing dynamic range (part 2)

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image processing apparatus 2 Image input means 3 Correction | amendment image creation means 4 Several correction value calculation means 5 Dynamic range compression means 7 Image output part

Claims (5)

Image input means for inputting a color image composed of pixels having a plurality of color information;
A corrected image creating means for creating a corrected image by correcting the density of color information of pixels of the color image so as to increase the brightness of the color image;
Among the pixels in the corrected image, with respect to a pixel having color information with a density value exceeding a predetermined dynamic range on the low density side, the pixel so that all the color information of the pixel falls within the dynamic range. Correction value calculation means for calculating a correction value for correcting all the color information of
Dynamic range compression means for applying dynamic range compression processing to the pixels having color information of density values exceeding the dynamic range among the pixels of the corrected image using the correction values obtained from the correction value calculation means; Prepared,
Non-identical correction candidates in which the correction value calculation means calculates different correction candidate values for each color information so that all color information of pixels having color information of density values exceeding the dynamic range falls within the dynamic range A skin color pixel having a skin color ratio r representing a skin color-likeness defined in advance in a predetermined color space is a skin color pixel having a value calculation unit and having color information of density values exceeding the dynamic range. In this case, the image processing apparatus is characterized in that the correction candidate value calculated by the non-identical correction candidate value calculation unit is the correction value. The same correction value calculation means calculates correction candidate values having the same value for each color information so that all color information of pixels having color information of density values exceeding the dynamic range fall within the dynamic range. suggestions value calculator further includes a, the case where a pixel with color information of the density values exceeding the dynamic range is the skin color index r is a non-skin color pixels is 0 ≦ r <1, the skin color ratio of the pixel The correction candidate value of the non-identical correction candidate value calculation unit and the correction candidate value of the same correction candidate value calculation unit are set so that the ratio of the correction candidate value of the same correction candidate value calculation unit increases as The image processing apparatus according to claim 1, wherein the correction value is calculated by weighted averaging.   3. The image processing apparatus according to claim 1, wherein the corrected image creating means performs correction by adding a constant value to all color information of each pixel. An image input step of inputting a color image composed of pixels having a plurality of color information;
A corrected image creating step of creating a corrected image in which the density of color information of pixels of the color image is corrected so as to increase the brightness of the color image;
Among the pixels in the corrected image, with respect to a pixel having color information with a density value exceeding a predetermined dynamic range on the low density side, the pixel so that all the color information of the pixel falls within the dynamic range. A correction value calculating step for calculating a correction value for correcting all of the color information;
A dynamic range compression step of performing dynamic range compression processing using a correction value obtained from the correction value calculation step on a pixel having color information of a density value exceeding the dynamic range among the pixels of the corrected image. Prepared,
Non-identical correction candidates in which the correction value calculation step calculates different correction candidate values for each color information so that all color information of pixels having color information of density values exceeding the dynamic range are within the dynamic range A skin color pixel having a skin color ratio r representing a skin color likeness defined in advance in a predetermined color space , wherein the pixel having color information having a density value exceeding the dynamic range has a value calculation step. A correction candidate value calculated in the non-identical correction candidate value calculation step is used as the correction value. Computer
Image input means for inputting a color image composed of pixels having a plurality of color information;
A corrected image creating means for creating a corrected image by correcting the density of color information of pixels of the color image so as to increase the brightness of the color image;
Among the pixels in the corrected image, with respect to a pixel having color information with a density value exceeding a predetermined dynamic range on the low density side, the pixel so that all the color information of the pixel falls within the dynamic range. Correction value calculation means for calculating a correction value for correcting all the color information of
Functions as a dynamic range compression unit that performs dynamic range compression processing using a correction value obtained from the correction value calculation unit on a pixel having color information of a density value exceeding the dynamic range among the pixels of the corrected image. A program to
Non-identical correction candidates in which the correction value calculation means calculates different correction candidate values for each color information so that all color information of pixels having color information of density values exceeding the dynamic range falls within the dynamic range A skin color pixel having a skin color ratio r representing a skin color-likeness defined in advance in a predetermined color space is a skin color pixel having a value calculation unit and having color information of density values exceeding the dynamic range. In this case, the correction candidate value calculated by the non-identical correction candidate value calculation unit is used as the correction value.