JP2005079623A - Method, apparatus and program of correcting white balance - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve the effect of white balance of a color image. <P>SOLUTION: In a white balance correcting method, the image signal of each pixel of a color image is multiplied by predetermined coefficients, a group of gray candidate pixels corresponding to each of the coefficients is obtained using a pixel included near a black body locus curve of the gray as a gray candidate pixel, the coefficient at which the number of the pixels of the group of gray candidate pixels is the maximum is used as the optimum coefficient, the color temperature of the photographing light source of the color image is obtained from the color temperature of each pixel contained in the group of gray candidate pixels corresponding to the optimum coefficient, and the image signal of the color image which is multiplied by the optimum coefficient is corrected by the difference between the obtained color temperature and the color temperature of reference white. In this method, the sum obtained by weighted addition of the number of counts of each of the gray candidate pixels for the group of gray candidate images using a weighting coefficient shown in the figure is used as the number of pixels of the group of gray candidate pixels. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a method and apparatus for performing white balance correction on a color image, and a program therefor.

  For digital images obtained by photoelectrically reading photographic images recorded on photographic films such as negative films and color reversal films with a reading device such as a scanner, and images obtained by taking images with a digital still camera (DSC) Thus, various image processing is performed for printing. As one of these image processes, in the case of a color image, there is a white balance correction process for avoiding the deterioration of the print image quality caused by the color of the photographing light source being reflected in the print.

  In addition, digital images obtained by reading color photographic images recorded on photographic film and color images obtained by DSC (hereinafter abbreviated as all images) are displayed on a monitor, not limited to printing. It is used in various forms. Even when the image is displayed on the monitor, it is necessary to remove the influence of the color of the photographing light source to obtain a beautiful image with white balance.

  Conventionally, as a representative method of white balance correction, a LATD (Large Area Transmission Density) method based on the Evans principle (hypothesis) that “all colors in the world are averaged to become gray” is used. A correction method is used. In this correction method, correction is performed so that the average color of the corrected image is gray.

  However, since the LATD method white balance correction method cannot distinguish whether the color applied to the image is caused by the photographic light source or the subject, it works well when the cause is the photographic light source. When the cause is the subject, there is a problem that the image quality of the corrected image is deteriorated. For example, in the case of an image obtained by taking a picture of a woman wearing red clothes, when white balance is applied using the LATD method, the red clothes become cloudy and the female's complexion color deteriorates in order to make the entire image gray. End up.

  In addition, most recent DSCs are equipped with an AWB (Auto White Balance) function that automatically performs white balance correction on an image obtained by shooting, and an image output from the DSC is the WAB function. The white balance is corrected by. The white balance correction processing by the AWB function in the DSC is a simple operation that operates to uniformly multiply R, G, and B of the image signals of all the pixels immediately after shooting, and when it works well. In some cases, it does not function, and it is impossible to make all images into a white balance image. Therefore, it is necessary to remove the influence when the AWB function does not work well.

  In order to solve the above-described problem, Patent Document 1 detects pixels having a predetermined color (for example, gray or skin color) suitable for obtaining a color temperature of a light source for an image to be processed. A method has been proposed in which the color temperature of a photographic light source is obtained from the color temperature of the pixel and white balance correction is performed based on the obtained color temperature of the photographic light source. Note that this method can be applied to images obtained by reading a color negative film or images obtained by DSC. Hereinafter, gray pixels are detected from an image obtained by DSC. The principle of this method will be described using white balance correction as an example. For example, consider a case where a scene including a gray portion (gray and its approximate color) is photographed with natural daylight with a color temperature of 4000 K by DSC. At this time, the image signal (R, G, B) of the photographed gray portion is converted into chromaticity coordinates (r, b) by the following equation (1) and plotted on the chromaticity diagram.

r = R / (R + G + B)
b = B / (R + G + B) (1)

FIG. 1 shows a chromaticity diagram. In FIG. 1, a curve Gy is a gray black body locus curve. The black body locus is the color temperature T, the black body radiant energy distribution of the color temperature T is P (λ), the spectral reflectance distribution of the subject is ρ (λ), and the spectral sensitivity distribution of the DSC CCD sensor is Si (λ ) (I = R, G, B), Ei calculated by the following equation (2) is converted into chromaticity coordinates (r, b) by the above equation (1) and plotted on the chromaticity diagram. This is the locus when the color temperature T is moved. A black body locus exists for each spectral distribution of the CCD sensor and the color of the subject, and a gray black body locus is obtained as a spectral reflectance ρ (λ) 1 of the subject when the subject is gray.

Ei = ∫P (λ) ρ (λ) Si (λ) dλ (2)

When the image signal of the gray portion in the scene described above is plotted on the chromaticity diagram shown in FIG. 1, it is considered that the image signal is scattered in the vicinity Gy0 of 4000K of the gray black body locus Gy. However, as described above, since most recent DSCs have an AWB (auto white balance) function, when this function works well, the gray part is in the vicinity of standard white (for example, 5500K) GY1. If this function does not work well, it will be sprayed to a location whose location is unknown from the color temperature 4000K of the photographic light source (for example, location A shown in FIG. 1).

  Therefore, a process of converting a gray part with an unknown position, such as A shown in FIG. 1, included in the DSC image into a reference white neighborhood Gy <b> 1 is a white balance correction process, and this process is performed for all pixels in the image. If applied, an image with white balance can be obtained. However, since the position of the portion A shown in FIG. 1 is unknown, it is impossible to directly convert A to the reference white neighborhood Gy1. Therefore, in the method proposed in Patent Document 1, the white balance processing is divided into two stages of conversion using vectors α and β shown in FIG. As shown in the figure, the vector α is a vector for correcting the amount of deviation from the black body locus Gy caused by the imperfection of the AWB function on the DSC side, and the portion A shown in FIG. It is converted into a part Gy0 on the black body locus Gy. Further, the vector β is a vector for converting the portion of Gy0 on the black body locus into the neighborhood Gy1 of the reference white (for example, 5500 K). By combining these two vectors α and β, the reference A is obtained from A shown in FIG. Conversion to the white neighborhood Gy1 is realized.

  Of the two vectors α and β, it is difficult to obtain a vector α that moves the portion A shown in FIG. 1 to Gy0 on the black body locus Gy. If Gy0 on the black body locus is obtained, the color temperature T of the photographing light source can be obtained from Gy0. Then, it is easy to obtain a vector β representing a conversion for transferring this Gy0 (color temperature T) to Gy1 (color temperature 5500K) on the black body locus Gy.

  On the other hand, the AWB function of the DSC acts so that the image signals R, G, and B of all the pixels immediately after photographing are uniformly multiplied by a constant. As a result, if the image signal deviates from the black body locus, if the operation is reversed, the vicinity of the color temperature of the imaging light source (4000K in this example) on the black body locus (Gy0 portion in FIG. 1) ), Many gray parts should be scattered. Since the AWB function of the DSC is the primary conversion for the image signal, the inverse conversion is also the primary conversion, that is, the DSC image signal R, G, B is multiplied by a predetermined coefficient (in FIG. 1, The image signals R, G, and B are converted into chromaticity coordinates). In the method proposed in Patent Document 1, the vector α is obtained using this. Specifically, the DSC image signal is multiplied by a predetermined coefficient, and the multiplied signal is compared with a gray black body locus for each coefficient. Assuming that the detected pixel has a high possibility of gray, it is determined as a gray candidate pixel. The determination of whether or not it is in the vicinity may be made based on whether or not the distance is within a range of 0.01 on the chromaticity coordinates (r, b).

  Thus, a gray candidate pixel group composed of gray candidate pixels is obtained for each coefficient to be multiplied. The number of pixels included in these gray candidate pixel groups is counted, and the coefficient corresponding to the gray candidate pixel group having the largest number of gray candidate pixels (corresponding to the portion A shown in FIG. 1) is the optimum coefficient, that is, A portion A shown in FIG. 1 can be used as a vector α for converting Gy0 on the black body locus.

  If the signal of each pixel of the gray candidate pixel group corresponding to this vector α is converted into Gy0 on the black body locus Gy shown in FIG. 1, the average value of the color temperature of each pixel of Gy0 is obtained. The color temperature of the photographic light source can be obtained. Since the vector β for converting the obtained color temperature into the reference white can be obtained, the white balance correction process is performed by applying the vectors α and β to all the pixels of the image. .

  Further, when obtaining the vector α, not only gray, but also a skin color with a high frequency of photographing and a little color conversion depending on the type may be used in the subject. The skin color seems to vary considerably depending on race (white, yellow, black, etc.), but according to the measured spectrum, the difference is mainly brightness, and the shape of the spectrum does not change much. Since the change in taste is small, the skin color can be used for color identification. That is, using the skin color black body locus instead of the gray body locus, the skin color candidate pixel group is obtained in the same manner as the gray candidate pixel group described above, and the skin color candidate pixel having the largest number of skin color candidate pixels is obtained. The vector α may be obtained by using the coefficient corresponding to the group as the optimum coefficient.

  Further, in order to obtain the vector α more accurately, both the gray candidate pixel group and the skin color candidate pixel group are detected, and the number of gray candidate pixels included in the gray candidate pixel group and the skin color candidate pixel group are included. The coefficient having the largest sum with the number of skin color candidate pixels may be the optimum coefficient, that is, the vector α.

  In addition, when a scene including a gray portion and a skin color portion is shot with a uniform light source, the color temperature of the shooting light source obtained using gray and the color temperature of the shooting light source obtained using skin color should match. By utilizing this, the coefficient having the smallest difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group may be obtained as the optimum coefficient, that is, the vector α.

  Further, the sum of the number of gray candidate pixels included in the gray candidate pixel group and the number of skin color candidate pixels included in the skin color candidate pixel group is the largest, and the average color temperature of the gray candidate pixel group and the average of the skin color candidate pixel group The coefficient having the smallest difference from the color temperature may be set as the optimum coefficient, and the accuracy of the vector α may be further increased.

Thus, in the method of obtaining the vector α and the vector β and applying the white balance correction to the image signals of all the pixels in the image, true gray and / or skin color pixels in the image are obtained. Since the white balance correction is performed by detecting the color temperature of the photographic light source, it is possible to obtain a more appropriate white balance correction effect than the white balance correction method based on the LATD method, and to the variation in the AWB function of the DSC. Is not affected.
JP 2002-152772 A

  However, in the method described in Patent Document 1, when obtaining the optimum coefficient, that is, the vector α, the number of gray candidate pixels in the gray candidate pixel group is obtained by treating the count number of each gray candidate pixel equally. Yes. That is, for each gray candidate pixel group, one gray candidate pixel is counted as one to obtain the number of gray candidate pixels included in each gray candidate pixel group, and this gray candidate pixel group corresponding to the largest number of gray candidate pixels is obtained. The coefficient is the optimum coefficient. The same applies when skin color is used.

  Also, when obtaining the vector α using the coefficient having the smallest difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group as the optimum coefficient, the color temperature of each candidate pixel is handled equally. Thus, the average color temperature of the gray candidate pixel group and the skin color candidate pixel group is obtained.

  On the other hand, even if the candidate pixel (gray candidate pixel or skin color candidate pixel) included in the same candidate pixel group (gray candidate pixel group or skin color candidate pixel group), the probability of true gray or true skin color is different. Considering the probability that candidate pixels included in the same candidate pixel group are true gray or true skin color, it is desired to improve the accuracy of the optimum coefficient and realize better white balance correction.

  In view of the above circumstances, the present invention obtains a vector α that is an optimum coefficient in consideration of the probability that candidate pixels included in the same candidate pixel group are true gray or true skin color, and can obtain a better effect. An object of the present invention is to provide a white balance correction method and apparatus, and a program therefor.

The first white balance correction method of the present invention multiplies an image signal of each pixel of a color image by a predetermined coefficient, and as a result, a pixel that enters the vicinity of a gray black body locus curve is determined as a gray candidate pixel. A gray candidate pixel group corresponding to each coefficient is obtained, and the coefficient that maximizes the number of pixels of the gray candidate pixel group is set as an optimal coefficient, and the gray candidate pixel group corresponding to the optimal coefficient is set as a gray pixel group. The color temperature of the photographing light source of the color image is obtained from the color temperature of each pixel included in the gray pixel group, and the image signal of the color image multiplied by the optimum coefficient is used as the obtained color temperature and the reference. In the white balance correction method that corrects only the difference from the white color temperature,
The gray candidate pixel group is a sum obtained by weighting and adding the count number of each gray candidate pixel for each gray candidate image group so that the weight of the count number becomes higher as the pixel is closer to the gray black body locus curve. The number of pixels is the characteristic.

  That is, the first white balance correction method of the present invention is to perform white balance correction by obtaining the color temperature of the photographing light source using gray, and it should be positioned on the gray black body locus curve. The gray candidate pixel group having the largest number of pixels (that is, the gray pixel group) is used as the optimum coefficient for converting the gray portion shifted from the gray black body locus curve back to the gray black body locus curve. The point of obtaining the color temperature of the photographic light source by obtaining the corresponding coefficient is the same as the method described in Patent Document 1, but was described in Patent Document 1 when obtaining the number of pixels of each gray candidate pixel group. As in the method, the gray candidate pixels included in each gray candidate pixel group are simply counted and added, that is, one gray candidate pixel is counted as 1, and the sum of the count numbers is applicable. Rather than using the number of pixels in the gray candidate pixel group, even a gray candidate pixel included in the same gray candidate pixel group may be a true gray pixel as the pixel is closer to the gray black body locus curve. Focusing on the fact that the gray candidate pixel is closer to the gray black body locus curve, the sum obtained by weighting and adding the count number of each gray candidate pixel for each gray candidate pixel group so that the weight of the count number is higher. Is the number of pixels of the gray candidate pixel group. That is, among the gray candidate pixels included in the same gray candidate group, the number of counts (one) for the gray candidate pixels located on the gray black body locus curve (the distance from the black body locus curve is zero). The gray candidate pixel is such that the gray candidate pixel that is farther from the gray black body locus curve has a smaller weight than the maximum weight. The sum obtained by weighting and adding the count number of gray candidate pixels for each group is set as the number of pixels of the gray candidate pixel group.

  In addition, “a pixel entering the vicinity of the black body locus curve” means that the image signal of the pixel (here, a signal multiplied by the predetermined coefficient) is a distance between the position in the chromaticity diagram and the black body locus curve. Means a pixel that is less than or equal to the value (for example, 0.03). The “distance” may be a value obtained by any method as long as the distance between the pixel and the black body locus curve can be represented (however, all of the same gray candidate pixel group). For example, the length of the normal line from the position of the pixel to the black body locus curve (for example, in the case of the pixel Q shown in FIG. 1, the length of the normal line) Of course, the length of the line drawn from the position of the pixel to the black body locus curve along the direction parallel to any one of the chromaticity coordinates (for example, the pixel Q shown in FIG. 1) In this case, the length L2) of the line parallel to the chromaticity coordinate r may be used. As a result of an actual experiment, the length of a line drawn from the pixel position to the black body locus curve along the direction parallel to the chromaticity coordinate r is used as the distance between the pixel position and the black body locus curve. Since the best white balance effect is obtained, the length of the line drawn from the pixel position to the black body locus curve along the direction parallel to the chromaticity coordinate r is defined as the pixel position and the black body locus curve. It is preferable to use it as a distance.

  Further, the color temperature of the gray pixel included in the gray pixel group means the color temperature in the gray black body locus curve of the pixel. For example, in the case of the pixel Q shown in FIG. The color temperature value of the point Q1 on the gray black body locus curve Gy is obtained.

  Further, the “color temperature” in the present invention is not limited to the color temperature itself, and any numerical value may be used as long as it can represent the color temperature. For example, the usual numerical value represented directly by the unit K is used, but of course, the unit is not K, but the color temperature can be expressed while being indirectly, mired shown in the following formula (3) A value may be used.

mired = 10 ⁶ / T (3)
T: Color temperature value (K)

The description of “pixels in the vicinity of the black body locus” and “color temperature” in the above description is applied to all the white balance correction methods, apparatuses, and programs of the present invention, and the description is omitted in the following description. To do.

In the second white balance correction method of the present invention, the image signal of each pixel of the color image is multiplied by a predetermined coefficient, and as a result, a pixel that enters the vicinity of the gray black body locus curve is determined as a gray candidate pixel. A gray candidate pixel group corresponding to each coefficient is obtained, and the coefficient that maximizes the number of pixels of the gray candidate pixel group is set as an optimal coefficient, and the gray candidate pixel group corresponding to the optimal coefficient is set as a gray pixel group. The color temperature of the photographing light source of the color image is obtained from the color temperature of each pixel included in the gray pixel group, and the image signal of the color image multiplied by the optimum coefficient is used as the obtained color temperature and the reference. In the white balance correction method that corrects only the difference from the white color temperature,
The sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of gray pixels in the sample color image to the total number of pixels of the sample color image Correction coefficient according to the presence frequency of the gray candidate pixel group, the gray candidate pixel group for each gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels of the color image. A value obtained by correcting the number is used as the number of pixels of the gray candidate pixel group.

  That is, in the second white balance correction method of the present invention, as in the first white balance correction method, the color temperature of the photographic light source is obtained using gray and white balance correction is performed. The optimal coefficient for performing the conversion to return the gray portion deviated from the gray black body locus curve to the gray body locus curve, although it should be located on the body locus curve, has the largest number of pixels. The coefficient corresponding to many gray candidate pixel groups (that is, gray pixel groups) is obtained to obtain the color temperature of the photographing light source. However, when obtaining the number of pixels of each gray candidate pixel group, it differs from the first white balance correction method. A value obtained by correcting the number of gray candidate pixels included in the gray candidate pixel group from the viewpoint is set as the number of pixels of the gray candidate pixel group. The details will be described below.

  As the ratio of the number of gray candidate pixels included in the gray candidate group detected from the color image to the total number of pixels in the color image increases, the gray candidate pixel included in the gray candidate pixel group becomes a true gray pixel. Is likely. Based on this tendency, when obtaining the number of pixels of the gray candidate pixel group, a correction coefficient that increases as the ratio increases according to the ratio of the gray candidate pixel group. If the value obtained by correcting the number is set as the number of pixels of the gray candidate pixel group, the optimum coefficient should be obtained with higher accuracy.

  On the other hand, FIG. 2A shows a sample color image in which the number of gray pixels in the sample color image is a large number (tens of thousands) of color images (the sample color image) obtained by photographing various scenes. The ratio of the total number of pixels (all pixels) to each image is calculated for each image, and the result of determining the existence frequency of sample color images having the ratio for each ratio (for example, 20,000 sample color images) If there are 200 sample color images having the above-mentioned ratio of w, the presence frequency of the sample color image corresponding to the above-mentioned ratio of w is 200 / 20,000, that is, 1%). It is a figure which shows the relationship between the obtained said ratio and the presence frequency of a sample color image. As shown in the figure, the greater the ratio of the number of gray pixels to the total number of pixels, the lower the existence frequency of the sample color image having the ratio, but the decrease rate of the existence frequency decreases as the ratio increases. Become.

  Since there is such a statistical tendency, the above-mentioned “correction coefficient that increases as the ratio increases” uses this correction coefficient that reflects the statistical tendency shown in FIG. The optimum coefficient can be obtained with better accuracy than using a correction coefficient that does not reflect a general tendency.

  FIG. 2B shows an example of the correction coefficient reflecting the statistical tendency shown in FIG. As shown in the figure, this correction coefficient is the above-mentioned “correction coefficient that increases as the ratio increases” and reflects the tendency shown in FIG. 2A, and the number of gray candidate pixels included in the gray candidate pixel group. Increases as the ratio increases, and the increase rate decreases as the ratio increases.

  The second white balance correction method of the present invention is a correction coefficient reflecting the statistical tendency shown in FIG. 2A (for example, the correction coefficient shown in FIG. 2B) and is included in the gray candidate pixel group. According to the ratio of the number of gray candidate pixels to the total number of pixels of the color image, the value obtained by correcting the number of gray candidate pixels for each gray candidate pixel group is the number of pixels in the gray candidate pixel group. And

  When obtaining the statistical tendency shown in FIG. 2A, it is desirable to use a pixel as close to a true gray pixel as possible as the gray pixel in the sample color image. For example, an optimal coefficient may be obtained by the method described in Patent Document 1, and a pixel included in a gray candidate pixel group corresponding to the optimal coefficient may be used, which is obtained using the first white balance correction method of the present invention. Pixels included in the gray candidate pixel group corresponding to the optimum coefficient may be used.

In the third white balance correction method of the present invention, the image signal of each pixel of the color image is multiplied by a predetermined coefficient, and as a result, a pixel that is in the vicinity of the skin color black body locus curve is defined as a skin color candidate pixel. A skin color candidate pixel group corresponding to each coefficient is obtained, the coefficient that maximizes the number of pixels of the skin color candidate pixel group is set as an optimal coefficient, and the skin color candidate pixel group corresponding to the optimal coefficient is set as a skin color pixel group. The color temperature of the photographing light source of the color image is obtained from the color temperature of each pixel included in the skin color pixel group, and the image signal of the color image multiplied by the optimum coefficient is used as the obtained color temperature and the reference. In the white balance correction method that corrects only the difference from the white color temperature,
For each skin color candidate pixel group, the sum obtained by weighting and adding the count number of each skin color candidate pixel so that the weight closer to the skin color black body locus curve is higher in the count number is added to the skin color candidate group. It is characterized by the number of pixels.

  The third white balance correction method of the present invention is the same as the first white balance correction method except that skin color is used instead of gray. That is, the third white balance correction method of the present invention performs white balance correction by obtaining the color temperature of the photographic light source using the skin color, and should be positioned on the skin color black body locus curve. The skin color candidate pixel group having the largest number of pixels (that is, the skin color pixel group) is used as the optimum coefficient for converting the skin color portion shifted from the skin color black body locus curve back to the skin color black body locus curve. The point of obtaining the color temperature of the photographing light source by obtaining the corresponding coefficient is the same as the method described in Patent Document 1, but it was described in Patent Document 1 when obtaining the number of pixels of each skin color candidate pixel group. As in the method, the flesh color candidate pixels included in each flesh color candidate pixel group are simply counted and added, that is, one flesh color candidate pixel is counted as 1, and the sum of the counts is calculated for the corresponding flesh color candidate pixel group. Number of pixels Rather than focusing on the fact that even skin color candidate pixels included in the same skin color candidate pixel group, pixels closer to the skin color black body locus curve are more likely to be true skin color pixels. Pixels of the skin color candidate pixel group are sums obtained by weighting and adding the count numbers of the skin color candidate pixels for each skin color candidate pixel group so that the skin color candidate pixels closer to the black body locus curve have higher count weights. It is a number. That is, among the skin color candidate pixels included in the same skin color candidate group, the count number (1) is determined for the skin color candidate pixels located on the skin color black body locus curve (the distance from the black body locus curve is zero). Skin color candidate pixels with the maximum weight assigned to the skin color candidate pixel, and the skin color candidate pixels that are farther away from the skin color black body locus curve have a smaller weight than the maximum weight. The sum obtained by weighting and adding the counts of the skin color candidate pixels for each group is set as the number of pixels of the skin color candidate pixel group.

The fourth white balance correction method of the present invention multiplies the image signal of each pixel of the color image by a predetermined coefficient, and as a result, a pixel that is in the vicinity of the skin color black body locus curve is used as the skin color candidate pixel. A skin color candidate pixel group corresponding to each coefficient is obtained, the coefficient that maximizes the number of pixels of the skin color candidate pixel group is set as an optimal coefficient, and the skin color candidate pixel group corresponding to the optimal coefficient is set as a skin color pixel group. The color temperature of the photographing light source of the color image is obtained from the color temperature of each pixel included in the skin color pixel group, and the image signal of the color image multiplied by the optimum coefficient is used as the obtained color temperature and the reference. In the white balance correction method that corrects only the difference from the white color temperature,
The sample color image for each of the proportions statistically determined using a number of the sample color images with respect to the proportion of the number of skin color pixels in the sample color image to the total number of pixels of the sample color image The correction coefficient according to the presence frequency of the flesh color candidate pixel group includes the flesh color candidate pixel group for each flesh color candidate pixel group according to the ratio of the number of flesh color candidate pixels included in the flesh color candidate pixel group to the total number of pixels of the color image. A value obtained by correcting the number is used as the number of pixels of the skin color candidate pixel group.

  That is, the fourth white balance correction method of the present invention performs white balance correction by obtaining the color temperature of the photographing light source using the skin color, as in the third white balance correction method. The optimal number of pixels is the most suitable coefficient for converting the skin-colored portion that deviates from the skin-colored black body locus curve into the skin-colored black body locus curve, although it should be located on the body locus curve. The color temperature of the photographing light source is obtained by obtaining coefficients corresponding to many skin color candidate pixel groups (that is, skin color pixel groups). However, when obtaining the number of pixels of each skin color candidate pixel group, it differs from the third white balance correction method. A value obtained by correcting the number of skin color candidate pixels included in the skin color candidate pixel group from the viewpoint is set as the number of pixels of the skin color candidate pixel group. The details will be described below.

  As the ratio of the number of skin color candidate pixels included in the skin color candidate group detected from the color image to the total number of pixels in the color image increases, the skin color candidate pixel included in the skin color candidate pixel group becomes a true skin color pixel. Is likely. Based on this tendency, when obtaining the number of pixels of the skin color candidate pixel group, according to the ratio of the skin color candidate pixel group, a correction coefficient that increases as the ratio increases and the skin color candidate pixels included in the skin color candidate pixel group If the value obtained by correcting the number is used as the number of pixels of the skin color candidate pixel group, the optimum coefficient should be obtained with higher accuracy.

  On the other hand, FIG. 3A shows a sample color image in which the number of skin color pixels in the sample color image is a large number (tens of thousands) of color images (the sample color image) obtained by photographing various scenes. The ratio of the total number of pixels (all pixels) to each image is calculated for each image, and the result of determining the existence frequency of sample color images having the ratio for each ratio (for example, 20,000 sample color images) If there are 200 sample color images having the above-mentioned ratio of w, the presence frequency of the sample color image corresponding to the above-mentioned ratio of w is 200 / 20,000, that is, 1%). It is a figure which shows the relationship between the obtained said ratio and the presence frequency of a sample color image. As shown in the figure, the greater the ratio of the number of skin color pixels to the total number of pixels, the lower the frequency of existence of the sample color image having that ratio (predetermined value, in the example of FIG. However, the rate of decrease in the existence frequency decreases as the ratio increases.

  Since there is such a statistical tendency, the above-mentioned “correction coefficient that increases as the ratio increases” uses this correction coefficient that reflects the statistical tendency shown in FIG. The optimum coefficient can be obtained with better accuracy than using a correction coefficient that does not reflect a general tendency.

  FIG. 3B shows an example of a correction coefficient that reflects the statistical tendency shown in FIG. As shown in the figure, this correction coefficient is the aforementioned “correction coefficient that increases as the ratio increases”, and reflects the tendency shown in FIG. 3A, and the number of skin color candidate pixels included in the skin color candidate pixel group. Depending on the ratio of the total number of pixels, the larger the ratio is, the larger the ratio is, and the larger the ratio is, the smaller the increase rate is. After the predetermined value, the larger the ratio, the smaller.

  The fourth white balance correction method of the present invention is a correction coefficient that reflects the statistical tendency shown in FIG. 3A (for example, the correction coefficient shown in FIG. 3B) and is included in the skin color candidate pixel group. According to the ratio of the number of skin color candidate pixels to the total number of pixels in the color image, the value obtained by correcting the number of skin color candidate pixels for each skin color candidate pixel group is the number of pixels in the skin color candidate pixel group. And

  When obtaining the statistical tendency shown in FIG. 3A, it is desirable to use a pixel that is as close as possible to the true skin color pixel as the skin color pixel in the sample color image. For example, an optimum coefficient may be obtained by the method described in Patent Document 1, and a pixel included in a skin color candidate pixel group corresponding to the optimum coefficient may be used, which is obtained by using the third white balance correction method of the present invention. Alternatively, pixels included in the flesh color candidate pixel group corresponding to the optimum coefficient may be used.

The fifth white balance correction method of the present invention uses both gray and skin color in order to obtain the color temperature with higher accuracy, and “the sum of the number of pixels of the gray candidate pixel group and the number of pixels of the skin color candidate group”. The gray candidate pixel group and the skin color candidate pixel group corresponding to the coefficient with the largest is a gray pixel group and a skin color pixel group, and performs white balance correction. Specifically, in the fifth white balance correction method of the present invention, the image signal of each pixel of the color image is multiplied by a predetermined coefficient, and as a result, pixels that fall in the vicinity of the gray black body locus curve are obtained. Obtaining a gray candidate pixel group corresponding to each coefficient as a gray candidate pixel, and a skin color candidate pixel group corresponding to each coefficient as a skin color candidate pixel that is in the vicinity of a skin color black body locus curve; The coefficient that maximizes the sum of the number of pixels in the group and the number of pixels in the skin color candidate pixel group is set as an optimal coefficient, and the gray candidate pixel group and the skin color candidate pixel group corresponding to the optimal coefficient are gray pixels. A color temperature of the photographing light source of the color image is obtained from the color temperature of each pixel included in the gray pixel group and the color temperature of each pixel included in the skin color pixel group. Is multiplied An image signal of the color image, the white balance correction method for correcting only the difference between the determined color temperature and reference white,
Sum obtained by weighting and adding the count number of each gray candidate pixel for each gray candidate image group so that the weight of the count number becomes higher as the pixel is closer to the gray black body locus curve, or a sample color image The presence frequency of the sample color image for each of the proportions statistically obtained using a number of the sample color images with respect to the proportion of the number of gray pixels in the total number of pixels of the sample color image. The number of gray candidate pixels included in the gray candidate pixel group is corrected for each gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels in the color image. The value obtained as a result, the number of pixels of the gray candidate pixel group, and / or
A sum obtained by weighting and adding the count number of each skin color candidate pixel for each skin color candidate pixel group so that the weight of the count number becomes higher as the pixel closer to the skin color black body locus curve, or a sample color image The presence frequency of the sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of skin color pixels to the total number of pixels of the sample color image The number of flesh color candidate pixels included in the flesh color candidate pixel group is corrected for each flesh color candidate pixel group according to the proportion of the total number of pixels in the color image. The value obtained in this way is used as the number of pixels of the skin color candidate pixel group.

The sixth white balance correction method of the present invention uses both gray and skin color, and corresponds to the coefficient having the smallest “difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group”. The gray candidate pixel group and the skin color candidate pixel group to be subjected to white balance correction are based on becoming a gray pixel group and a skin color pixel group. Specifically, in the sixth white balance correction method of the present invention, the image signal of each pixel of the color image is multiplied by a predetermined coefficient, and as a result, pixels that fall in the vicinity of the gray black body locus curve are obtained. Obtaining a gray candidate pixel group corresponding to each coefficient as a gray candidate pixel, and a skin color candidate pixel group corresponding to each coefficient as a skin color candidate pixel that is in the vicinity of a skin color black body locus curve; The coefficient that minimizes the difference between the average color temperature of the group and the average color temperature of the skin color candidate pixel group is set as an optimal coefficient, and the gray candidate pixel group and the skin color candidate pixel group corresponding to the optimal coefficient are gray. The color temperature of the photographing light source of the color image is obtained from the color temperature of each pixel included in the gray pixel group and the color temperature of each pixel included in the skin color pixel group. Coefficient is An image signal calculated by said color image, the white balance correction method for correcting only the difference between the determined color temperature and reference white,
A value obtained by performing a weighted average operation on the color temperature of each gray candidate pixel included in the gray candidate pixel group so that the gray candidate pixel closer to the gray black body locus curve has a higher color temperature weight. A step of setting an average color temperature of a gray candidate pixel group, and / or
A value obtained by performing a weighted average operation on the color temperature of each skin color candidate pixel included in the skin color candidate pixel group so that the skin temperature candidate pixel closer to the skin color black body locus curve has a higher color temperature weight. It has the process of setting it as the average color temperature of a skin color candidate pixel group.

The seventh white balance correction method of the present invention uses both gray and skin color, has the largest “sum of the number of pixels of the gray candidate pixel group and the number of skin color candidate pixels”, and the “average of the gray candidate pixel group” A gray candidate pixel group and a skin color candidate pixel group corresponding to a coefficient having a small difference between the color temperature and the average color temperature of the skin color candidate pixel group are a gray pixel group and a skin color pixel group, and performs white balance correction. It is. Specifically, in the seventh white balance correction method of the present invention, the image signal of each pixel of the color image is multiplied by a predetermined coefficient, and as a result, pixels that enter the vicinity of the gray black body locus curve are obtained. A gray candidate pixel group corresponding to each coefficient as a gray candidate pixel, and a skin color candidate pixel group corresponding to each coefficient as a skin color candidate pixel that is in the vicinity of a skin color black body locus curve are obtained, and the gray candidate pixel The coefficient that maximizes the sum of the number of pixels of the group and the number of pixels of the skin color candidate pixel group, and minimizes the difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group And the gray candidate pixel group and the skin color candidate pixel group corresponding to the optimal coefficient as a gray pixel group and a skin color pixel group, respectively, and the color temperature and the skin color of each pixel included in the gray pixel group Included in pixel group The color temperature of the color image photographing light source is obtained from the color temperature of each pixel to be corrected, and the image signal of the color image multiplied by the optimum coefficient is corrected by the difference between the obtained color temperature and the reference white color. In the white balance correction method,
Sum obtained by weighting and adding the count number of each gray candidate pixel for each gray candidate image group so that the weight of the count number becomes higher as the pixel is closer to the gray black body locus curve, or a sample color image The presence frequency of the sample color image for each of the proportions statistically obtained using a number of the sample color images with respect to the proportion of the number of gray pixels in the total number of pixels of the sample color image. The number of gray candidate pixels included in the gray candidate pixel group is corrected for each gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels in the color image. The value obtained as a result, the number of pixels of the gray candidate pixel group, and / or
Sum obtained by weighting and adding the count number of each skin color candidate pixel for each skin color candidate pixel group so that the weight of the count number becomes higher as the pixel closer to the flesh color black body locus curve, or a sample color image The presence frequency of the sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of skin color pixels to the total number of pixels of the sample color image The number of flesh color candidate pixels included in the flesh color candidate pixel group is corrected for each flesh color candidate pixel group for each flesh color candidate pixel group according to the ratio of the total number of pixels in the color image. And / or a step of setting the obtained value as the number of pixels of the skin color candidate pixel group, and / or
A value obtained by performing a weighted average operation on the color temperature of each gray candidate pixel included in the gray candidate pixel group so that the gray candidate pixel closer to the gray black body locus curve has a higher color temperature weight. A step of setting an average color temperature of a gray candidate pixel group, and / or
A value obtained by performing a weighted average operation on the color temperature of each skin color candidate pixel included in the skin color candidate pixel group so that the skin temperature candidate pixel closer to the skin color black body locus curve has a higher color temperature weight. It has the process of setting it as the average color temperature of a skin color candidate pixel group.

  In the first to seventh white balance correction methods of the present invention, the gray candidate pixel group and / or the skin color candidate pixel group may be acquired from all the pixels of the color image. Of these pixels, only the pixels effective for obtaining the gray candidate pixel group and / or the skin color candidate pixel group are used, and the gray candidate pixel group and / or the skin color candidate pixel group are acquired from these effective pixels. It is preferable to do so. For example, when gray is used as in the first white balance correction method, the image signal of all the pixels of the color image is multiplied by a predetermined coefficient, and as a result, pixels that enter the vicinity of the gray black body locus curve Instead of obtaining a gray candidate pixel group corresponding to each coefficient with a gray candidate pixel as a gray candidate pixel, it is highly likely that the pixel is an effective pixel for obtaining a gray candidate pixel among all the pixels of the color image, that is, a gray candidate pixel. Pixels are extracted, and the extracted image signals of these pixels are multiplied by a predetermined coefficient. As a result of the multiplication, pixels that fall in the vicinity of the gray blackbody locus curve are gray candidates and gray corresponding to each coefficient. It is preferable to obtain a candidate pixel group.

  As an effective pixel for obtaining the gray candidate pixel group and the skin color candidate pixel group, for example, a high-saturation pixel (for example, a pixel whose distance from the gray axis of R = G = B is a predetermined threshold or more in a color image. As for the threshold value, the same threshold value may be used for each hue, or the threshold value may be changed depending on the hue), a pixel that is located in a bright area in the image, or a combination of these pixels It can be.

  When skin color is used, a face part in a color image may be extracted by a conventionally known face extraction technique, and the face part pixel may be an effective pixel.

  Further, in the case of a color image acquired by DSC, since the color temperature of the photographing light source is not correctly reflected in the highest pixel (the pixel whose pixel value is reduced due to the saturation of the CCD), such a pixel. If the color temperature of the photographic light source is determined using, the accuracy will be reduced and the white balance correction will be adversely affected.Therefore, pixels up to a predetermined percentage of the highlights are regarded as invalid pixels, and these invalid pixels are removed. The middle pixel may be an effective pixel.

  According to the first to seventh white balance correction methods of the present invention, the photographic light source is detected from the color temperature of each pixel included in the candidate pixel group (gray candidate pixel group and / or skin color candidate pixel group) corresponding to the optimum coefficient. When obtaining the color temperature, the closer to the black body locus curve (the gray black body locus curve in the case of the gray candidate pixel group, the skin color black body locus curve in the case of the skin color candidate pixel group), the weight of the color temperature. It is preferable that an average color temperature obtained by performing a weighted average calculation of the color temperatures of the pixels included in the candidate pixel group corresponding to the optimal coefficient is obtained as the color temperature of the photographing light source so that the color temperature becomes higher.

  In the above white balance correction method, the black body locus curve is used. However, because the daylight locus curve is substantially the same as the black body locus curve, in all the above white balance correction methods, the black body locus curve is used. A daylight locus curve may be used instead of the curve.

A first white balance correction apparatus of the present invention includes a multiplying unit that multiplies an image signal of each pixel of a color image by a predetermined coefficient;
As a result of the multiplication, a pixel that falls in the vicinity of a gray black body locus curve is used as a gray candidate pixel to obtain a gray candidate pixel group corresponding to each coefficient, and the number of gray candidate pixel groups for obtaining the number of gray candidate pixel groups Acquisition means;
Optimal coefficient acquisition means for acquiring, as an optimal coefficient, the coefficient corresponding to the gray candidate pixel group having the maximum number of pixels acquired by the gray candidate pixel group pixel number acquisition means;
The gray candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition unit is set as a gray pixel group, and the color temperature of the photographing light source of the color image is determined from the color temperature of each pixel included in the gray pixel group. A light source color temperature acquisition means to be obtained;
The image signal of the color image is multiplied by the optimum coefficient, and the multiplied image signal is corrected by the difference between the color temperature obtained by the light source color temperature acquisition unit and the color temperature of the reference white. And a white balance correction device having correction means for
The gray candidate pixel group pixel number acquisition unit weights and adds the count number of each gray candidate pixel for each gray candidate image group so that the weight closer to the gray black body locus curve is higher in the count number. The sum obtained in this way is the number of pixels of the gray candidate pixel group.

The second white balance correction apparatus of the present invention includes a multiplying unit that multiplies the image signal of each pixel of the color image by a predetermined coefficient,
As a result of the multiplication, a pixel that falls in the vicinity of a gray black body locus curve is used as a gray candidate pixel to obtain a gray candidate pixel group corresponding to each coefficient, and the number of gray candidate pixel groups for obtaining the number of gray candidate pixel groups Acquisition means;
Optimal coefficient acquisition means for acquiring, as an optimal coefficient, the coefficient corresponding to the gray candidate pixel group having the maximum number of pixels acquired by the gray candidate pixel group pixel number acquisition means;
The gray candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition unit is set as a gray pixel group, and the color temperature of the photographing light source of the color image is determined from the color temperature of each pixel included in the gray pixel group. A light source color temperature acquisition means to be obtained;
The image signal of the color image is multiplied by the optimum coefficient, and the multiplied image signal is corrected by the difference between the color temperature obtained by the light source color temperature acquisition unit and the color temperature of the reference white. And a white balance correction device having correction means for
The gray candidate pixel group pixel number acquisition means statistically obtains the ratio of the number of gray pixels in the sample color image to the total number of pixels in the sample color image using a large number of sample color images. The correction coefficient according to the frequency of existence of the sample color image for each of the ratios is determined according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels of the color image. A value obtained by correcting the number of gray candidate pixels for each gray candidate pixel group is set as the number of pixels of the gray candidate pixel group.

A third white balance correction apparatus according to the present invention includes a multiplying unit that multiplies an image signal of each pixel of a color image by a predetermined coefficient;
As a result of the multiplication, a pixel that enters the vicinity of the skin color black body locus curve is used as a skin color candidate pixel to obtain a skin color candidate pixel group corresponding to each of the coefficients, and a skin color candidate pixel group pixel number for obtaining the number of pixels of the skin color candidate pixel group Acquisition means;
Optimal coefficient acquisition means for acquiring, as an optimal coefficient, the coefficient corresponding to the skin color candidate pixel group in which the number of pixels acquired by the skin color candidate pixel group pixel number acquisition means is maximized;
The skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition unit is defined as a skin color pixel group, and the color temperature of the photographing light source of the color image is determined from the color temperature of each pixel included in the skin color pixel group. A light source color temperature acquisition means to be obtained;
The image signal of the color image is multiplied by the optimum coefficient, and the multiplied image signal is corrected by the difference between the color temperature obtained by the light source color temperature acquisition unit and the color temperature of the reference white. And a white balance correction device having correction means for
The skin color candidate pixel group pixel number acquisition unit weights and adds the count number of each skin color candidate pixel for each skin color candidate pixel group so that a pixel closer to the skin color black body locus curve has a higher weight of the count number. The sum obtained in this way is used as the number of pixels of the skin color candidate pixel group.

A fourth white balance correction apparatus according to the present invention includes a multiplying unit that multiplies an image signal of each pixel of a color image by a predetermined coefficient;
As a result of the multiplication, a pixel that enters the vicinity of the skin color black body locus curve is used as a skin color candidate pixel to obtain a skin color candidate pixel group corresponding to each of the coefficients, and a skin color candidate pixel group pixel number for obtaining the number of pixels of the skin color candidate pixel group Acquisition means;
Optimal coefficient acquisition means for acquiring, as an optimal coefficient, the coefficient corresponding to the skin color candidate pixel group in which the number of pixels acquired by the skin color candidate pixel group pixel number acquisition means is maximized;
The skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition unit is defined as a skin color pixel group, and the color temperature of the photographing light source of the color image is determined from the color temperature of each pixel included in the skin color pixel group. A light source color temperature acquisition means to be obtained;
The image signal of the color image is multiplied by the optimum coefficient, and the multiplied image signal is corrected by the difference between the color temperature obtained by the light source color temperature acquisition unit and the color temperature of the reference white. And a white balance correction device having correction means for
The skin color candidate pixel group pixel number acquisition means statistically obtains the ratio of the number of skin color pixels in the sample color image to the total number of pixels of the sample color image using a large number of sample color images. The correction coefficient corresponding to the frequency of existence of the sample color image for each of the ratios, and the number of skin color candidate pixels included in the skin color candidate pixel group according to the ratio of the total number of pixels of the color image A value obtained by correcting the number of skin color candidate pixels for each skin color candidate pixel group is used as the number of pixels of the skin color candidate pixel group.

A fifth white balance correction apparatus of the present invention includes a multiplying unit that multiplies an image signal of each pixel of a color image by a predetermined coefficient,
As a result of the multiplication, pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are defined as skin color candidate pixels. A candidate pixel group pixel number acquisition unit for obtaining a skin color candidate pixel group corresponding to a coefficient and calculating the number of pixels of the gray candidate pixel group and the number of pixels of the skin color candidate pixel group;
Optimal coefficient acquisition means for acquiring, as an optimal coefficient, the coefficient that maximizes the sum of the number of pixels of the gray candidate pixel group acquired by the candidate pixel group pixel number acquisition means and the number of pixels of the skin color candidate pixel group;
The gray candidate pixel group and the skin color candidate pixel group corresponding to the appropriate coefficient acquired by the optimal coefficient acquisition unit are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and Light source color temperature obtaining means for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction unit that multiplies the image signal of the color image by the optimum coefficient and corrects the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition unit and a reference white color. In a white balance correction device having
The candidate pixel group pixel number acquisition unit weights and adds the count number of each gray candidate pixel for each gray candidate image group so that the weight closer to the gray black body locus curve is higher in the count number. Or the ratio of the number of gray pixels in the sample color image to the total number of pixels in the sample color image obtained statistically using a number of sample color images. The gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels of the color image, with a correction coefficient according to the frequency of existence of the sample color image for each ratio. A value obtained by correcting the number of gray candidate pixels for each of the gray candidate pixel groups, and / or
In a sample color image, or a sum obtained by weighting and adding each skin color candidate pixel count number for each skin color candidate pixel group so that a weight closer to the skin color black body locus curve is higher in the count number The ratio of the number of skin color pixels to the total number of pixels of the sample color image is the frequency of existence of the sample color image for each of the ratios statistically determined using a number of the sample color images. The number of skin color candidate pixels included in the skin color candidate pixel group is corrected for each skin color candidate pixel group for each skin color candidate pixel group according to the ratio of the number of skin color candidate pixels included in the skin color candidate pixel group to the total number of pixels in the color image. The value obtained in this way is used as the number of pixels of the skin color candidate pixel group.

A sixth white balance correction apparatus according to the present invention includes a multiplying unit that multiplies an image signal of each pixel of a color image by a predetermined coefficient;
As a result of the multiplication, pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are defined as skin color candidate pixels. A candidate color group obtaining means for obtaining a skin color candidate pixel group corresponding to a coefficient, and obtaining an average color temperature of the gray candidate pixel group and an average color temperature of the skin color candidate pixel group;
An optimum coefficient obtaining unit that obtains, as an optimum coefficient, the coefficient that minimizes the difference between the average color temperature of the gray candidate pixel group obtained from the candidate pixel group average color temperature and the average color temperature of the skin color candidate pixel group; ,
The gray candidate pixel group and the skin color candidate pixel group corresponding to the appropriate coefficient acquired by the optimal coefficient acquisition unit are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and Light source color temperature obtaining means for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction unit that multiplies the image signal of the color image by the optimum coefficient and corrects the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition unit and a reference white color. In a white balance correction device having
The candidate pixel group average color temperature acquisition unit is configured so that a gray candidate pixel closer to the gray black body locus curve has a higher color temperature weight so that the color temperature of each gray candidate pixel included in the gray candidate pixel group is higher. A value obtained by performing a weighted average calculation on the gray candidate pixel group as an average color temperature, and / or
A value obtained by performing a weighted average operation on the color temperature of each skin color candidate pixel included in the skin color candidate pixel group so that the skin temperature candidate pixel closer to the skin color black body locus curve has a higher color temperature weight. The processing is performed to obtain the average color temperature of the skin color candidate pixel group.

A seventh white balance correction apparatus of the present invention includes a multiplying unit for multiplying an image signal of each pixel of a color image by a predetermined coefficient,
As a result of the multiplication, pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are each defined as the skin color candidate pixel. Candidate pixel group obtaining means for obtaining a skin color candidate pixel group corresponding to the coefficient;
Candidate pixel group pixel number obtaining means for obtaining the number of pixels of the gray candidate pixel group and the number of pixels of the skin color candidate pixel group obtained by the candidate pixel group obtaining means;
Candidate pixel group average color temperature acquisition means for obtaining an average color temperature of the gray candidate pixel group acquired by the candidate pixel group acquisition means and an average color temperature of the skin color candidate pixel group;
The sum of the number of pixels of the gray candidate pixel group acquired by the candidate pixel group pixel number acquisition unit and the number of pixels of the skin color candidate pixel group is maximized, and acquired by the candidate pixel group average color temperature acquisition unit An optimum coefficient obtaining unit that obtains, as an optimum coefficient, the coefficient that minimizes the difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group;
The gray candidate pixel group and the skin color candidate pixel group corresponding to the appropriate coefficient acquired by the optimal coefficient acquisition unit are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and Light source color temperature obtaining means for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction unit that multiplies the image signal of the color image by the optimum coefficient and corrects the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition unit and a reference white color. In a white balance correction device having
The candidate pixel group pixel number acquisition unit weights and adds the count number of each gray candidate pixel for each gray candidate image group so that the weight closer to the gray black body locus curve is higher in the count number. Or the ratio of the number of gray pixels in the sample color image to the total number of pixels in the sample color image obtained statistically using a number of sample color images. The gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels of the color image, with a correction coefficient according to the frequency of existence of the sample color image for each ratio. A value obtained by correcting the number of gray candidate pixels for each of the gray candidate pixel groups, and / or
Sum obtained by weighting and adding the count number of each skin color candidate pixel for each skin color candidate pixel group so that the weight of the count number becomes higher as the pixel closer to the flesh color black body locus curve, or a sample color image The presence frequency of the sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of skin color pixels to the total number of pixels of the sample color image The number of flesh color candidate pixels included in the flesh color candidate pixel group is corrected for each flesh color candidate pixel group for each flesh color candidate pixel group according to the ratio of the total number of pixels in the color image. The value obtained in this way is used as the number of pixels of the skin color candidate pixel group.

An eighth white balance correction apparatus according to the present invention includes a multiplying unit that multiplies the image signal of each pixel of a color image by a predetermined coefficient,
As a result of the multiplication, pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are each defined as the skin color candidate pixel. Candidate pixel group obtaining means for obtaining a skin color candidate pixel group corresponding to the coefficient;
Candidate pixel group pixel number obtaining means for obtaining the number of pixels of the gray candidate pixel group and the number of pixels of the skin color candidate pixel group obtained by the candidate pixel group obtaining means;
Candidate pixel group average color temperature acquisition means for obtaining an average color temperature of the gray candidate pixel group acquired by the candidate pixel group acquisition means and an average color temperature of the skin color candidate pixel group;
The sum of the number of pixels of the gray candidate pixel group acquired by the candidate pixel group pixel number acquisition unit and the number of pixels of the skin color candidate pixel group is maximized, and acquired by the candidate pixel group average color temperature acquisition unit An optimum coefficient obtaining unit that obtains, as an optimum coefficient, the coefficient that minimizes the difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group;
The gray candidate pixel group and the skin color candidate pixel group corresponding to the appropriate coefficient acquired by the optimal coefficient acquisition unit are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and Light source color temperature obtaining means for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction unit that multiplies the image signal of the color image by the optimum coefficient and corrects the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition unit and a reference white color. In a white balance correction device having
The candidate pixel group average color temperature acquisition unit is configured so that a gray candidate pixel closer to the gray black body locus curve has a higher color temperature weight so that the color temperature of each gray candidate pixel included in the gray candidate pixel group is higher. A value obtained by performing a weighted average calculation on the gray candidate pixel group as an average color temperature, and / or
A value obtained by performing a weighted average operation on the color temperature of each skin color candidate pixel included in the skin color candidate pixel group so that the skin temperature candidate pixel closer to the skin color black body locus curve has a higher color temperature weight. The processing is performed to obtain the average color temperature of the skin color candidate pixel group.

  The first to eighth white balance correction devices of the present invention extract effective pixels for obtaining the gray candidate pixel group and / or skin color candidate pixel group from the pixels of the color image, and It is preferable that the image processing apparatus further includes an effective pixel extraction unit that uses only pixels for obtaining the gray candidate pixel group and / or the skin color candidate pixel group.

  In the white balance device of the present invention, the light source color temperature acquisition unit includes each pixel included in the candidate pixel group corresponding to the optimal coefficient so that a pixel closer to the black body locus curve has a higher color temperature weight. It is preferable that an average color temperature obtained by weighted average calculation of the color temperature of the pixel is obtained as the color temperature of the photographing light source.

  In the white balance correction device of the present invention, a daylight locus curve may be used instead of the black body locus curve.

A first program according to the present invention includes a multiplication process for multiplying an image signal of each pixel of a color image by a predetermined coefficient;
As a result of the multiplication, a pixel that enters the vicinity of the gray black body locus curve is regarded as a gray candidate pixel, a gray candidate pixel group corresponding to each coefficient is obtained, and the number of gray candidate pixel groups for obtaining the number of pixels in the gray candidate pixel group Acquisition process,
An optimal coefficient acquisition process for acquiring, as an optimal coefficient, the coefficient corresponding to the gray candidate pixel group in which the number of pixels acquired by the gray candidate pixel group pixel number acquisition process is maximized;
The gray candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process is defined as a gray pixel group, and the color temperature of the imaging light source of the color image is determined from the color temperature of each pixel included in the gray pixel group. The desired light source color temperature acquisition process;
The image signal of the color image is multiplied by the optimum coefficient, and the multiplied image signal is corrected by the difference between the color temperature obtained by the light source color temperature acquisition process and the color temperature of the reference white. A program for causing a computer to perform correction processing,
In the gray candidate pixel group pixel number acquisition process, the count number of each gray candidate pixel is weighted and added for each gray candidate image group such that the weight closer to the gray black body locus curve is higher in the count number. The sum obtained in this way is the number of pixels of the gray candidate pixel group.

A second program of the present invention includes a multiplication process for multiplying an image signal of each pixel of a color image by a predetermined coefficient,
As a result of the multiplication, a pixel that enters the vicinity of the gray black body locus curve is regarded as a gray candidate pixel, a gray candidate pixel group corresponding to each coefficient is obtained, and the number of gray candidate pixel groups for obtaining the number of pixels in the gray candidate pixel group Acquisition process,
An optimal coefficient acquisition process for acquiring, as an optimal coefficient, the coefficient corresponding to the gray candidate pixel group in which the number of pixels acquired by the gray candidate pixel group pixel number acquisition process is maximized;
The gray candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process is defined as a gray pixel group, and the color temperature of the imaging light source of the color image is determined from the color temperature of each pixel included in the gray pixel group. The desired light source color temperature acquisition process;
The image signal of the color image is multiplied by the optimum coefficient, and the multiplied image signal is corrected by the difference between the color temperature obtained by the light source color temperature acquisition process and the color temperature of the reference white. A program for causing a computer to perform correction processing,
The gray candidate pixel group pixel number acquisition process is statistically obtained using a large number of sample color images with respect to the ratio of the number of gray pixels in the sample color image to the total number of pixels in the sample color image. The correction coefficient according to the frequency of existence of the sample color image for each of the ratios, and the number of gray candidate pixels included in the gray candidate pixel group according to the ratio of the total number of pixels of the color image according to the ratio A value obtained by correcting the number of gray candidate pixels for each gray candidate pixel group is set as the number of pixels of the gray candidate pixel group.

A third program of the present invention includes a multiplication process for multiplying an image signal of each pixel of a color image by a predetermined coefficient,
As a result of the multiplication, a pixel that enters the vicinity of the skin color black body locus curve is used as a skin color candidate pixel to obtain a skin color candidate pixel group corresponding to each of the coefficients, and a skin color candidate pixel group pixel number for obtaining the number of pixels of the skin color candidate pixel group Acquisition process,
An optimal coefficient acquisition process for acquiring the coefficient corresponding to the skin color candidate pixel group that maximizes the number of pixels acquired by the skin color candidate pixel group pixel number acquisition process;
The skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process is defined as a skin color pixel group, and the color temperature of the photographing light source of the color image is determined from the color temperature of each pixel included in the skin color pixel group. The desired light source color temperature acquisition process;
The image signal of the color image is multiplied by the optimum coefficient, and the multiplied image signal is corrected by the difference between the color temperature obtained by the light source color temperature acquisition process and the color temperature of the reference white. A program for causing a computer to execute correction processing to be performed,
In the skin color candidate pixel group pixel number acquisition process, the count number of each skin color candidate pixel is weighted and added for each skin color candidate pixel group so that the weight closer to the skin color black body locus curve is higher in the count number. The sum obtained in this way is used as the number of pixels of the skin color candidate pixel group.

A fourth program of the present invention includes a multiplication process for multiplying an image signal of each pixel of a color image by a predetermined coefficient,
As a result of the multiplication, a pixel that enters the vicinity of the skin color black body locus curve is used as a skin color candidate pixel to obtain a skin color candidate pixel group corresponding to each of the coefficients, and a skin color candidate pixel group pixel number for obtaining the number of pixels of the skin color candidate pixel group Acquisition process,
An optimal coefficient acquisition process for acquiring the coefficient corresponding to the skin color candidate pixel group that maximizes the number of pixels acquired by the skin color candidate pixel group pixel number acquisition process;
The skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process is defined as a skin color pixel group, and the color temperature of the photographing light source of the color image is determined from the color temperature of each pixel included in the skin color pixel group. The desired light source color temperature acquisition process;
The image signal of the color image is multiplied by the optimum coefficient, and the multiplied image signal is corrected by the difference between the color temperature obtained by the light source color temperature acquisition process and the color temperature of the reference white. A program for causing a computer to execute correction processing to be performed,
The skin color candidate pixel group pixel number acquisition process is statistically obtained using a large number of sample color images with respect to the ratio of the number of skin color pixels in the sample color image to the total number of pixels in the sample color image. The correction coefficient according to the frequency of existence of the sample color image for each of the ratios is determined according to the ratio of the number of skin color candidate pixels included in the skin color candidate pixel group to the total number of pixels of the color image. A value obtained by correcting the number of skin color candidate pixels for each skin color candidate pixel group is used as the number of pixels of the corresponding skin color candidate pixel group.

A fifth program of the present invention includes a multiplication process for multiplying an image signal of each pixel of a color image by a predetermined coefficient,
As a result of the multiplication, pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are defined as skin color candidate pixels. Obtaining a skin color candidate pixel group corresponding to a coefficient, obtaining a pixel number of the gray candidate pixel group and a number of pixels of the skin color candidate pixel group, and obtaining a pixel number of candidate pixel groups,
An optimum coefficient obtaining process for obtaining, as an optimum coefficient, the coefficient that maximizes the sum of the number of pixels of the gray candidate pixel group obtained by the candidate pixel group pixel number obtaining process and the number of pixels of the skin color candidate pixel group;
The gray candidate pixel group and the skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and A light source color temperature acquisition process for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction process for multiplying the image signal of the color image by the optimum coefficient, and correcting the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition process and a reference white color Is a program that causes a computer to execute
In the candidate pixel group pixel number acquisition processing, the count number of each gray candidate pixel is weighted and added for each gray candidate image group so that the weight closer to the gray black body locus curve is higher in the count number. Or the ratio of the number of gray pixels in the sample color image to the total number of pixels in the sample color image, statistically determined using a number of sample color images. The gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels of the color image with a correction coefficient according to the frequency of existence of the sample color image for each ratio. A procedure in which the value obtained by correcting the number of gray candidate pixels for each time is the number of pixels of the gray candidate pixel group, and / or
Sum obtained by weighting and adding the count number of each skin color candidate pixel for each skin color candidate pixel group so that the weight of the count number becomes higher as the pixel closer to the flesh color black body locus curve, or a sample color image The presence frequency of the sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of skin color pixels to the total number of pixels of the sample color image The number of flesh color candidate pixels included in the flesh color candidate pixel group is corrected for each flesh color candidate pixel group according to the proportion of the total number of pixels in the color image. The value obtained in this manner is used as the number of pixels of the skin color candidate pixel group.

A sixth program of the present invention includes a multiplication process for multiplying an image signal of each pixel of a color image by a predetermined coefficient,
As a result of the multiplication, pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are defined as skin color candidate pixels. Obtaining a skin color candidate pixel group corresponding to a coefficient, and obtaining an average color temperature of the gray candidate pixel group and an average color temperature of the skin color candidate pixel group;
An optimal coefficient acquisition process for acquiring, as an optimal coefficient, the coefficient that minimizes the difference between the average color temperature of the gray candidate pixel group acquired by the candidate pixel group average color temperature and the average color temperature of the skin color candidate pixel group; ,
The gray candidate pixel group and the skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and A light source color temperature acquisition process for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction process for multiplying the image signal of the color image by the optimum coefficient, and correcting the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition process and a reference white color Is a program that causes a computer to execute
In the candidate pixel group average color temperature acquisition process, the color temperature of each gray candidate pixel included in the gray candidate pixel group is such that a gray candidate pixel closer to the gray black body locus curve has a higher color temperature weight. The procedure of setting the value obtained by performing the weighted average calculation as the average color temperature of the gray candidate pixel group, and / or
A value obtained by performing a weighted average operation on the color temperature of each skin color candidate pixel included in the skin color candidate pixel group so that the skin temperature candidate pixel closer to the skin color black body locus curve has a higher color temperature weight. It has the procedure which makes it the average color temperature of a skin color candidate pixel group.

A seventh program of the present invention includes a multiplication process for multiplying an image signal of each pixel of a color image by a predetermined coefficient,
As a result of the multiplication, the pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are each defined as the skin color candidate pixel. A candidate pixel group acquisition process for obtaining a skin color candidate pixel group corresponding to the coefficient;
A candidate pixel group pixel number acquisition process for obtaining the number of pixels of the gray candidate pixel group acquired by the candidate pixel group acquisition process and the number of pixels of the skin color candidate pixel group;
A candidate pixel group average color temperature acquisition process for obtaining an average color temperature of the gray candidate pixel group acquired by the candidate pixel group acquisition process and an average color temperature of the skin color candidate pixel group;
The sum of the number of pixels of the gray candidate pixel group acquired by the candidate pixel group pixel number acquisition process and the number of pixels of the skin color candidate pixel group is maximized, and acquired by the candidate pixel group average color temperature acquisition process An optimal coefficient acquisition process for acquiring, as an optimal coefficient, the coefficient that minimizes the difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group;
The gray candidate pixel group and the skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and A light source color temperature acquisition process for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction process for multiplying the image signal of the color image by the optimum coefficient, and correcting the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition process and a reference white color Is a program that causes a computer to execute
The candidate pixel group pixel number acquisition processing weights and adds the count number of each gray candidate pixel for each gray candidate image group so that the weight closer to the gray black body locus curve is higher in the count number. Or the ratio of the number of gray pixels in the sample color image to the total number of pixels in the sample color image obtained statistically using a number of sample color images. The gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels of the color image, with a correction coefficient according to the frequency of existence of the sample color image for each ratio. A value obtained by correcting the number of gray candidate pixels for each step, and / or the number of pixels in the gray candidate pixel group, and / or
Sum obtained by weighting and adding the count number of each skin color candidate pixel for each skin color candidate pixel group so that the weight of the count number becomes higher as the pixel closer to the flesh color black body locus curve, or a sample color image The presence frequency of the sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of skin color pixels to the total number of pixels of the sample color image The number of flesh color candidate pixels included in the flesh color candidate pixel group is corrected for each flesh color candidate pixel group according to the proportion of the total number of pixels in the color image. The value obtained in this manner is used as the number of pixels of the skin color candidate pixel group.

An eighth program of the present invention includes a multiplication process for multiplying an image signal of each pixel of a color image by a predetermined coefficient;
As a result of the multiplication, the pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are each defined as the skin color candidate pixel. A candidate pixel group acquisition process for obtaining a skin color candidate pixel group corresponding to the coefficient;
A candidate pixel group pixel number acquisition process for obtaining the number of pixels of the gray candidate pixel group acquired by the candidate pixel group acquisition process and the number of pixels of the skin color candidate pixel group;
A candidate pixel group average color temperature acquisition process for obtaining an average color temperature of the gray candidate pixel group acquired by the candidate pixel group acquisition process and an average color temperature of the skin color candidate pixel group;
The sum of the number of pixels of the gray candidate pixel group acquired by the candidate pixel group pixel number acquisition process and the number of pixels of the skin color candidate pixel group is maximized, and acquired by the candidate pixel group average color temperature acquisition process An optimal coefficient acquisition process for acquiring, as an optimal coefficient, the coefficient that minimizes the difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group;
The gray candidate pixel group and the skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and A light source color temperature acquisition process for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction process for multiplying the image signal of the color image by the optimum coefficient, and correcting the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition process and a reference white color Is a program that causes a computer to execute
In the candidate pixel group average color temperature acquisition process, the color temperature of each gray candidate pixel included in the gray candidate pixel group is such that a gray candidate pixel closer to the gray black body locus curve has a higher color temperature weight. The procedure of setting the value obtained by performing the weighted average calculation as the average color temperature of the gray candidate pixel group, and / or
A value obtained by performing a weighted average operation on the color temperature of each skin color candidate pixel included in the skin color candidate pixel group so that the skin temperature candidate pixel closer to the skin color black body locus curve has a higher color temperature weight. It has the procedure which makes it the average color temperature of a skin color candidate pixel group.

  The first to eighth programs of the present invention extract effective pixels for obtaining the gray candidate pixel group and / or skin color candidate pixel group from the pixels of the color image, and only the effective pixels are extracted. It is preferable that the effective pixel extraction process for obtaining the gray candidate pixel group and / or the skin color candidate pixel group is further performed by a computer.

  According to the program of the present invention, as the light source color temperature acquisition process, each pixel included in the candidate pixel group corresponding to the optimal coefficient is weighted so that a pixel closer to the black body locus curve has a higher color temperature weight. It is preferable that an average color temperature obtained by weighted average calculation of the color temperature is obtained as the color temperature of the photographing light source.

  In each process executed by the program of the present invention, a daylight locus curve may be used instead of the black body locus curve.

  According to the white balance correction method and apparatus and the program therefor according to the present invention, when obtaining the number of gray candidate pixel groups and / or skin color candidate pixel groups, the candidate pixel count number (one candidate pixel has a count number). The probability that a gray candidate pixel is a true gray pixel and the skin color candidate pixel is a true flesh color, even if the candidate pixel is a candidate pixel included in the same candidate pixel group. Since the value obtained by weighting and adding the count number of candidate pixels according to the probability of becoming a pixel is used as the number of pixels of the candidate pixel group, the optimum coefficient obtained based on the number of pixels of the candidate pixel group and the optimum coefficient From the corresponding candidate pixel group, the color temperature of the photographing light source can be obtained with higher accuracy, and a better white balance correction effect can be obtained.

  According to the white balance correction method and apparatus and the program therefor according to the present invention, when determining the number of pixels of the gray candidate pixel group and / or the skin color candidate pixel group, a correction coefficient according to a statistically determined tendency (for example, The values obtained by correcting the sum obtained by simply adding the count numbers of the candidate pixels with the correction coefficients shown in FIGS. 2B and 3B corresponding to gray and skin color, respectively, are the pixels of the candidate pixel group. Therefore, the color temperature of the photographic light source can be obtained more accurately from the optimum coefficient obtained based on the number of pixels in the candidate pixel group and the candidate pixel group corresponding to the optimum coefficient, and a better white balance correction can be obtained. The effect of can be obtained.

  Based on the fact that the difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group should be the smallest, the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group Is obtained by simply averaging the color temperatures of the pixels included in the candidate pixel group as the color temperature of the candidate pixel group, even if it is a candidate pixel included in the same candidate pixel group. The value obtained by weighted averaging the color temperature of the candidate pixel according to the probability that the candidate pixel is a true gray pixel and the probability that the skin color candidate pixel is a true skin color pixel is used as the average color temperature of the candidate pixel group. Therefore, the optimum coefficient can be obtained with high accuracy, and consequently the color temperature of the photographing light source can be obtained with high accuracy.

  Also, when obtaining the color temperature of the photographic light source, the value obtained by performing a simple average operation on the color temperature of the candidate pixel group corresponding to the optimal coefficient is not used as the color temperature of the photographic light source, and the candidate pixel group corresponding to the optimal coefficient Since each candidate pixel included in is a value obtained by performing a weighted average calculation of the color temperature of each candidate pixel according to the probability of being a true gray pixel or a true skin color pixel, The color temperature of the photographing light source can be obtained with high accuracy, and the effect of good white balance correction can be achieved.

  In addition, when acquiring candidate pixels, effective pixels are extracted from all pixels in the color image, and candidate pixels are acquired from these effective pixels. The amount of calculation can be reduced and the color temperature of the photographic light source can be obtained with higher accuracy by removing invalid pixels such as the highest pixels that adversely affect the acquisition of the color temperature of the photographic light source. The effect of correction can be improved.

  Embodiments of the present invention will be described below with reference to the drawings.

  FIG. 4 is a block diagram showing a configuration of an image processing apparatus A which is an embodiment of the white balance correction method and apparatus of the present invention and a program therefor. Note that the image processing apparatus A according to the present embodiment performs white balance correction on a color image, and executes a white balance correction program read into an auxiliary storage device on a computer (for example, a personal computer). It is realized by. The white balance correction program is stored in an information storage medium such as a CDROM or distributed via a network such as the Internet and installed in a computer.

  As shown in the figure, the image processing apparatus A of the present embodiment performs first effective pixel extraction for extracting effective pixels from a color image S0 (R0, G0, B0) acquired by a digital still camera (DSC). The image signal corresponding to each coefficient by multiplying the image signals S1a and s1b of the effective pixels extracted by the means 10a and the second effective pixel extracting means 10b and the effective pixel extracting means 10a and 10b by a predetermined coefficient. And a gray candidate pixel group detecting unit for detecting a gray candidate pixel group S3a corresponding to each coefficient using the image signal group S2a obtained by the multiplication unit 15. 20a and a skin color candidate for detecting a skin color candidate pixel group S3b corresponding to each coefficient using the image signal groups S2a and S2b obtained by the multiplying unit 15 Pixel group detection means 20b, pixel number calculation means 30a for determining the number of pixels Ma of each gray candidate pixel group S3a and the number of pixels Mb of each skin color candidate pixel group S3b, and each gray candidate pixel group S3a The average color temperature calculating means 30b for determining the respective average color temperature Ta and the average color temperature Tb of each skin color candidate pixel group S3b, and the sum of the number of pixels of the gray candidate pixel group and the number of pixels of the skin color candidate pixel group Obtained by the optimum coefficient obtaining means 40 and the optimum coefficient obtaining means 40 for obtaining the coefficient having the smallest difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group as the optimum coefficient α. The light source color temperature acquisition means for obtaining the color temperature T of the photographing light source from the color temperature of each pixel included in the gray candidate pixel group S3a and the skin color candidate pixel group S3b corresponding to the optimum coefficient α. 50 and the optimum coefficient α obtained by the optimum coefficient obtaining means 40 and the color temperature of the photographing light source obtained by the light source color temperature obtaining means 50 are used to correct the image signals of all the pixels of the color image S0. Correction means 60 for obtaining corrected images S4 (R4, G4, B4).

  Next, each configuration of the image processing apparatus A shown in FIG. 1 will be described.

  The first effective pixel extracting means 10a first removes the following pixels in the color image S0 as invalid pixels.

1. Highly Saturated Pixel in Color Image Here, a pixel whose distance from the gray axis of R = G = B is a predetermined threshold or higher is defined as a highly saturated pixel. The same threshold value may be used for each hue, or the threshold value may be changed depending on the hue.

2. Highest pixel Here, the remaining pixels obtained by removing the invalid pixels by the first effective pixel extracting unit 10a, which uses the pixels up to a predetermined percentage (for example, 5%) of the highlight in the color image S0 as the highest pixel. The effective pixel S1a (R1a, G1a, B1a) is obtained by adding a pixel in the high luminance region in the color image S0, that is, a pixel having a luminance value equal to or higher than a predetermined threshold value to the pixel.

    The second effective pixel extraction means 10b is a face extraction means, extracts a face portion in the color image S0, and uses the extracted face portion pixels as effective pixels S1b (R1b, G1b, G1b). The effective pixel S1a extracted by 10a is added.

    The multiplication means 15 multiplies the image signals of the effective pixels S1a and S1b by a predetermined coefficient α0 (α1, α2) as shown in the following equation (4) to form a set of image signals corresponding to each coefficient α0. The image signal groups S2a and S2b are obtained, and the image signal group S2a is output to the gray candidate pixel group detection means 20a, and the image signal groups S2a and S2b are output to the skin color candidate pixel group detection means 20b. Note that it is not necessary to perform multiplication on the three signals R, G, and B, and it is sufficient if it can be performed on any two signals, here R and G.

R2a = α1R1a
G2a = α2G1a (4)
R2b = α1R1b
G2b = α2G1b

The gray candidate pixel group detection unit 20a and the skin color candidate pixel group detection unit 20b detect the gray candidate pixel group and the skin color candidate pixel group corresponding to each coefficient using the image signal group S2a, S2a, and S2b, respectively. Is. Hereinafter, the operations of the gray candidate pixel group detection unit 20a and the skin color candidate pixel group detection unit 20b will be described in detail with reference to FIG.

  In FIG. 1, Gy is a gray black body locus curve. As described above, the black body locus means that the color temperature is T, the black body radiant energy distribution at the color temperature T is P (λ), the spectral reflectance distribution of the subject is ρ (λ), and the spectral sensitivity of the DSC CCD sensor. When the distribution is Si (λ) (i = R, G, B), Ei calculated by the following equation (2) is converted into chromaticity coordinates (r, b) by equation (1), and chromaticity is obtained. It is a locus when the color temperature T is moved when plotted in the figure. A black body locus exists for each spectral distribution of the CCD sensor and the color of the subject, and a gray black body locus is obtained as the spectral reflectance ρ (λ) 1 of the subject when the subject is gray. The spectral sensitivity distribution Si (λ) of the CCD sensor is preferably unique to the DSC CCD sensor that captured the color image S0. However, the ideal spectral sensitivity distribution conforming to BT709 as shown in FIG. 5 is used. Also good.

r = R / (R + G + B)
b = B / (R + G + B) (1)


Ei = ∫P (λ) ρ (λ) Si (λ) dλ (2)

The skin color black body locus curve is obtained for the skin color in the same manner as the gray black body locus curve Gy, and is not shown here.

  For the image signal group S2a obtained by the multiplication means 15, the gray candidate pixel group detection means 20a first converts the (R, G, B) signal into chromaticity coordinates (r, b) according to the equation (1). The data is converted and plotted on the chromaticity diagram shown in FIG. Next, the gray candidate pixel group detection means 20a compares each signal of the image signal group S2a with the gray black body locus curve Gy, and has a pixel having a signal in the vicinity of the gray black body locus curve Gy (here, Compared with a point on the blackbody locus curve Gy, a pixel having a signal having a chromaticity coordinate r difference of 0.03 or less is detected as a gray candidate pixel. As described above, the gray candidate pixel group detection unit 20a detects gray candidate pixels for each coefficient α0, and detects a gray candidate pixel group S3a corresponding to each coefficient α0.

  The skin color candidate pixel group detection means 20b detects the skin color candidate pixel group S3b corresponding to each coefficient α0 using the image signals S2a and S2b, and instead of the gray black body locus curve Gy as the black body locus curve. Except for the point of using a skin-colored black body locus curve, the operation is substantially the same as that of the gray candidate pixel group detection unit 20a, and thus detailed description thereof is omitted here.

  In this way, gray candidate pixel groups S3a and S3b corresponding to the respective coefficients α0 (α1, α2) are detected by the gray candidate pixel group detection unit 20a and the skin color candidate pixel group detection unit 20b, respectively.

  The pixel number calculating unit 30a calculates the pixel number Ma of each gray candidate pixel group S3a and the pixel number Mb of each skin color candidate pixel group S3b. The average color temperature calculating unit 30b includes the gray candidate pixel group S3a. The average color temperature Ta and the average color temperature Tb of each skin color candidate pixel group S3b are calculated. First, the operation of the pixel number calculation means 30a will be described.

  The pixel number calculation means 30a weights the count number of each pixel of the gray candidate pixel group S3a with a weighting coefficient that increases the weight of the count number as the pixel is closer to the gray black body locus curve Gy shown in FIG. The sum obtained by the addition is set as the pixel number Ma of the gray candidate pixel group S3a. FIG. 6 shows the weighting coefficient (shown by a solid line) used by the pixel number calculation means 30a in the present embodiment. As shown in the figure, even if the pixels are included in the same gray candidate pixel group, if the pixel is on the gray black body locus curve Gy (the distance from the black body locus curve Gy is zero), the weighting coefficient for the count number is 1. In other words, when one pixel contributes 1 to the number of pixels in the pixel group, and the distance from the black body locus curve Gy is 0.02, the weighting coefficient for the count number Is 0.05, that is, one pixel contributes only 0.05 to the number of pixels in the pixel group. The dotted line in the figure indicates the weighting coefficient used when calculating the number of pixels of the candidate pixel group in the method proposed in Patent Document 1. As illustrated, in the method proposed in Patent Document 1, when calculating the number of pixels of the gray candidate pixel group, 1 is used as the weighting coefficient for all the pixels of the gray candidate pixel group, that is, the gray black One candidate pixel is supposed to contribute 1 pixel regardless of the distance from the body locus curve (the probability of being true gray).

  Similarly, the pixel number calculation means 30a weights the count number of each pixel of the skin color candidate pixel group S3b with a weighting coefficient that gives a higher weight to the count number of pixels closer to a skin color black body locus curve (not shown). The sum obtained by the addition is set as the number of pixels Mb of the skin color candidate pixel group S3b. Here, the pixel number calculation means 30a calculates the pixel number Mb of the flesh color candidate pixel group S3b using the weighting coefficient shown in FIG. 6 in the same manner as when calculating the pixel number of the gray candidate pixel group S3a.

  The average color temperature calculation unit 30b makes the weight of the color temperature higher for each pixel of the gray candidate pixel group S3a (for example, A portion shown in FIG. 1) as the pixel is closer to the gray black body locus curve Gy. A value obtained by performing a weighted average operation on the color temperature of each gray candidate pixel is set as the average color temperature Ta of the gray candidate pixel group S3a, and a skin-colored black body is applied to each pixel of the skin color candidate pixel group S3b. A value obtained by performing a weighted average calculation of the color temperature of each skin color candidate pixel so that the weight of the color temperature is higher for a pixel closer to the locus curve is defined as the average color temperature Tb of the skin color candidate pixel group S3b. Here, as an example, the average color temperature calculation unit 30b uses the weighting coefficient shown in FIG. 6 used by the pixel number calculation unit 30a as the color temperature weighting coefficient, and the average color temperature Ta of each candidate pixel group. And Tb are calculated.

  The pixel number Ma of each gray candidate pixel group S3a calculated by the pixel number calculation unit 30a and the pixel number Mb of the skin color candidate pixel group S3b, and the average color of each gray candidate pixel group S3a calculated by the average color temperature calculation unit 30b The temperature Ta and the average color temperature Tb of the skin color candidate pixel group S3b are output to the optimum coefficient acquisition unit 40. The optimum coefficient acquisition means 40 determines that “the sum of the number of pixels of the gray candidate pixel group and the skin color candidate pixel group is the largest” and “the difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color pixel group is The coefficient α0 is optimized by using “smallest” as an objective function to obtain the optimum coefficient α (α1, α2), which is output to the correction means 60 and the light source color temperature acquisition means 50. Here, the optimum coefficient α corresponds to the vector α shown in FIG.

  The light source color temperature acquisition means 50 estimates the color temperature of the photographing light source from the color temperatures of the gray candidate pixel group and the skin color candidate pixel group corresponding to the optimum coefficient α (α1, α2). The color temperature T of the photographing light source is obtained by further averaging the average color temperature T1a of the gray candidate pixel group corresponding to the optimum coefficient α and the average color temperature T1b of the skin color candidate pixel group corresponding to the optimum coefficient α. Is. The average color temperatures T1a and T1b of the gray candidate pixel group (gray pixel group) corresponding to the optimum coefficient α and the skin color candidate pixel group (skin color pixel group) corresponding to the optimum coefficient α may be newly obtained. Here, the light source color temperature acquisition unit acquires the average color temperatures T1a and T1b of the candidate pixel group corresponding to the optimum coefficient α from the average color temperature calculated by the average color temperature calculation unit 30b, and averages them. The value is calculated as the color temperature T of the light source. When averaging T1a and T1b, a value obtained by simple averaging as shown in the following equation (5) may be used as the color temperature T of the photographic light source. When gray color is important, T1a is used as the light source. The color temperature T may be set as follows. Furthermore, the color temperature T of the light source may be a value obtained by performing a weighted average operation on T1a and T1b according to the degree of emphasis on gray and skin color.

T = (T1a + T1b) / 2 (5)
Where T: color temperature of the photographic light source
T1a: Average color temperature of gray pixels
T1b: Average color temperature of skin color pixels

First, the correcting unit 60 obtains a conversion for converting the color temperature T acquired by the light source color temperature acquiring unit 50 into a reference white color (for example, 5500 K), that is, a vector β (β1, β2) shown in FIG.

  Using the optimum coefficients, that is, the vector α (α1, α2) and the vector β (β1, β2) shown in FIG. 1, correction processing is performed on the color image S0 (R0, G0, B0) according to the following equation (6). The corrected image S4 (R4, G4, B4) can be obtained.

R4 = α1β1R0
G4 = α2G0 (6)
B4 = β2B0
However,
R4, G4, B4: Pixel value of the corrected image S4 R0, G0, B0: Pixel value of the color image S0 (α1, α2): Optimal coefficient α

Further, since the conversion shown in the equation (6) includes not only the color balance but also the change in brightness, if the G signal is expressed as invariant, the following equation (7) is obtained.

R4 = (α1β1 / α2) R0
G4 = G0 (7)
B4 = (β2 / α2) B0
However,
R4, G4, B4: pixel values of the corrected image S4
R0, G0, B0: Pixel value of the color image S0
(Α1, α2): Optimal coefficient α

The correcting unit 60 performs the conversion shown in Expression (7) on the image signals of all the pixels of the color image S0, and obtains a corrected image S4.

  In the image processing apparatus A of the embodiment shown in FIG. 1, when the number of pixels of a candidate pixel group (gray candidate pixel group or skin color candidate pixel group) is obtained, and the optimum coefficient, that is, the vector α is obtained based on the number of pixels. Focusing on the fact that candidate pixels closer to the black body locus curve have a higher probability of becoming true gray pixels or true skin color pixels, weighting and adding the count number of each pixel according to the distance from the black body locus curve Since the vector α is obtained by using the obtained value as the number of pixels of the candidate pixel group, the accuracy of the color temperature of the photographing light source calculated based on the vector α can be improved, and the white balance is improved. The effect of correction can be achieved.

  FIG. 7 is a block diagram showing a configuration of an image processing apparatus B as an embodiment of the white balance correction method and apparatus and program of the present invention. It should be noted that the image processing apparatus B of the present embodiment is similar to the image processing apparatus A except that a pixel number calculating unit 30′a described later operates differently from the pixel number calculating unit 30a of the image processing apparatus A shown in FIG. Therefore, the same reference numerals are given to configurations that perform the same operations as the corresponding configurations of the image processing apparatus A, and the pixel number calculation unit 30′a of the image processing apparatus B has the same configuration. Only the operation will be described.

  As shown in FIG. 7, the image processing apparatus B according to the present embodiment has a first effective for extracting effective pixels from a color image S0 (R0, G0, B0) acquired by a digital still camera (DSC). The pixel extraction means 10a, the second effective pixel extraction means 10b, and the effective pixel image signals S1a and s1b extracted by the effective pixel extraction means 10a and 10b are multiplied by predetermined coefficients to correspond to the respective coefficients. Multiplication means 15 for obtaining image signal groups S2a and S2b as a set of image signals, and gray candidate pixel group for detecting gray candidate pixel group S3a corresponding to each coefficient using image signal group S2a obtained by multiplication means 15 Skin for detecting a skin color candidate pixel group S3b corresponding to each coefficient using the detection means 20a and the image signal groups S2a and S2b obtained by the multiplication means 15 A candidate pixel group detection unit 20b, a pixel number calculation unit 30'a for obtaining a pixel number M'a of each gray candidate pixel group S3a and a pixel number M'b of each skin color candidate pixel group S3b; Average color temperature calculation means 30b for determining the average color temperature Ta of each gray candidate pixel group S3a and the average color temperature Tb of each skin color candidate pixel group S3b, the number of pixels of the gray candidate pixel group and the skin color candidate pixels An optimum coefficient acquisition means 40 for acquiring, as an optimum coefficient α ′, a coefficient having the largest sum of the number of pixels of the group and the smallest difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group; The color temperature T ′ of the photographing light source is obtained from the color temperature of each pixel included in the gray candidate pixel group S3a and the skin color candidate pixel group S3b corresponding to the optimal coefficient α ′ acquired by the optimal coefficient acquisition unit 40. Using the source color temperature acquisition means 50, the optimum coefficient α ′ obtained by the optimum coefficient acquisition means 40, and the color temperature of the photographic light source obtained by the light source color temperature acquisition means 50, image signals of all pixels of the color image S0. And a correcting means 60 for correcting the image to obtain a corrected image S4 (R4, G4, B4).

  Here, only the operation of the pixel number calculating means 30'a will be described.

  The pixel number calculation means 30′a calculates the pixel number M′a of each gray candidate pixel group S3a and the pixel number M′b of each skin color candidate pixel group S3b. First, referring to FIG. The calculation of the pixel number M′a of the gray candidate pixel group S3a by the pixel number calculation unit 30′a will be described.

  As described above, FIG. 2B shows an example of the correction coefficient that reflects the statistical tendency shown in FIG. As shown in the figure, this correction coefficient is a “correction coefficient that increases as the ratio increases”, and reflects the tendency shown in FIG. 2A, so that the number of gray candidate pixels included in the gray candidate pixel group is all. Depending on the ratio of the number of pixels, the larger the ratio, the larger the ratio, and the larger the ratio, the smaller the increase rate. The pixel number calculation means 30′a of the image processing apparatus B shown in FIG. 7 calculates the value obtained by correcting the number of gray candidate pixels for each gray candidate pixel group S3a with the correction coefficient shown in FIG. It is assumed that the pixel number M′a of the gray candidate pixel group.

  Next, the calculation of the pixel number M′b of the flesh color candidate pixel group S3b by the pixel number calculation unit 30′a will be described with reference to FIG.

  As described above, FIG. 3B shows an example of the correction coefficient that reflects the statistical tendency shown in FIG. As shown in the figure, this correction coefficient is “a correction coefficient that increases as the ratio increases”, and reflects the tendency shown in FIG. 3A, and the number of skin color candidate pixels included in the skin color candidate pixel group is all. Depending on the ratio of the number of pixels, the larger the ratio is, the larger the ratio is, and the larger the ratio is, the smaller the increase rate is. After this value, the larger the ratio, the smaller. The number-of-pixels calculation unit 30′b of the image processing apparatus B of the present embodiment calculates the value obtained by correcting the number of skin color candidate pixels for each skin color candidate pixel group S3b with the correction coefficient shown in FIG. The number of pixels M′b of the skin color candidate pixel group is assumed.

  The optimum coefficient acquisition unit 40 of the image processing apparatus B calculates the average color temperature and the pixel number M′a of the gray candidate pixel group and the pixel number M′b of the skin color candidate pixel group obtained by the pixel number calculation unit 30′a. The optimum coefficient α ′ is obtained using the average color temperature Ta of the gray candidate pixel group and the average color temperature Tb of the skin color candidate pixel group obtained by the means 30b, and the light source color temperature acquisition means 50 corresponds to the optimum coefficient α ′. A value obtained by further averaging the average color temperatures T1′a and T1′b of the gray candidate pixel group and the skin color candidate pixel group to be calculated is calculated as the color temperature T ′ of the photographic light source. Using the vector β ′ corresponding to the vector β shown in FIG. 1 obtained based on the color temperature T ′ of the imaging light source acquired by the temperature acquisition unit 50 and the optimum coefficient α ′ acquired by the optimum coefficient acquisition unit 40, Correction processing for color image S0 Then, a corrected image S4 is obtained.

  As described above, the image processing apparatus B according to the present embodiment obtains the number of pixels of the candidate pixel group (gray candidate pixel group or skin color candidate pixel group), and obtains the optimum coefficient, that is, the vector α ′ based on the number of pixels. At this time, from a viewpoint different from that of the image processing apparatus A, a correction coefficient according to a statistically determined tendency according to the ratio of the number of candidate pixels included in the candidate pixel group to the total number of pixels of the color image ( The values obtained by correcting the sum obtained by simply adding the count numbers of the candidate pixels with the correction coefficients shown in FIGS. 2B and 3B corresponding to gray and skin color, respectively, are the pixels of the candidate pixel group. Therefore, the color temperature of the photographic light source can be obtained more accurately from the optimum coefficient obtained based on the number of pixels in the candidate pixel group and the candidate pixel group corresponding to the optimum coefficient, and a better white balance correction can be obtained. Can get the effect of That.

  The preferred embodiment of the present invention has been described above. However, the white balance correction method and apparatus of the present invention and the program therefor are not limited to the above-described embodiment, and various methods can be used without departing from the gist of the present invention. Increase, decrease, change can be added.

  For example, in the image processing apparatus A of the embodiment shown in FIG. 1 and the image processing apparatus B shown in FIG. 7, effective pixels are extracted from all the pixels of the color image S0, and these effective pixels are used as candidate pixels. The group detection, the optimum coefficient α (α ′) and the color temperature of the photographing light source are calculated, but the above-described processing is performed on the image obtained by reducing the color image S0, that is, the effective pixel. It is possible to reduce the amount of calculation by performing processing from extraction of the above to calculation of the optimum coefficient and the color temperature of the photographing light source, thereby speeding up the processing. In this case, if the thinning reduction process is used as the reduction process, the accuracy of the calculation can be improved. However, for example, the variation in the white balance correction effect between similar scenes such as a series of images obtained by continuous shooting by DSC can be reduced. From the viewpoint of reducing, it is desirable to use the average reduction process as the reduction process rather than the thinning reduction process.

  Further, the image processing apparatus A according to the embodiment shown in FIG. 1 and the image processing apparatus B shown in FIG. 7 extract effective pixels from the color image S0 in order to improve calculation accuracy and speed, Although calculation of the optimum coefficient and the color temperature of the light source is performed using pixels, the above-described processing may be performed using all pixels of the color image S0.

  Further, the image processing apparatus A of the embodiment shown in FIG. 1 and the image processing apparatus B shown in FIG. 7 use both gray and skin color, but use only one of gray and skin color. May be.

  Further, the image processing apparatus A according to the embodiment shown in FIG. 1 and the image processing apparatus B shown in FIG. 7 obtain the optimum coefficient by using both the number of pixels of the candidate pixel group and the average color temperature of the candidate pixel group. However, the optimum coefficient may be obtained using only one of the number of pixels and the average color temperature.

  The white balance correction method, apparatus, and program of the present invention are not limited to color images acquired by DSC, but are also applied to color images obtained by reading a photographic image acquired by a conventional negative color film with a reading device such as a scanner. Can be applied.

Chromaticity diagram for explaining the principle of the white balance method and apparatus and program of the present invention The figure which shows the tendency for which the relation of the existence frequency of a sample color image and the ratio which the gray pixel in a sample color image accounts to all the pixels was calculated statistically Correction coefficient for the number of gray candidate pixels reflecting the tendency shown in FIG. The figure which shows the tendency for which the relation of the existence frequency of a sample color image and the ratio which the skin color pixel in a sample color image occupies for all the pixels was calculated statistically Correction coefficient for the number of skin color candidate pixels reflecting the tendency shown in FIG. The block diagram which shows the structure of the image processing apparatus A used as embodiment of this invention. The figure which shows ideal spectral sensitivity distribution of the CCD sensor of BT709 conformity The figure for demonstrating operation | movement of the pixel number calculation means 30a and the average color temperature calculation means 30b in the image processing apparatus A shown in FIG. The block diagram which shows the structure of the image processing apparatus B used as embodiment of this invention

Explanation of symbols

DESCRIPTION OF SYMBOLS 10a 1st effective pixel extraction means 10b 2nd effective pixel extraction means 15 Multiplication means 20a Gray candidate pixel group detection means 20b Skin color candidate pixel group detection means 30a, 30'a Pixel number calculation means 30b Average color temperature calculation means 40 Optimal Coefficient acquisition means 50 Light source color temperature acquisition means 60 Correction means

Claims (26)

The image signal of each pixel of the color image is multiplied by a predetermined coefficient, and as a result, a gray candidate pixel group corresponding to each coefficient is obtained by setting a pixel that enters the vicinity of the gray black body locus curve as a gray candidate pixel. The coefficient that maximizes the number of pixels of the gray candidate pixel group is the optimum coefficient, the gray candidate pixel group corresponding to the optimum coefficient is the gray pixel group, and each pixel included in the gray pixel group A white balance that obtains a color temperature of the photographing light source of the color image from a color temperature and corrects the image signal of the color image multiplied by the optimum coefficient by a difference between the obtained color temperature and a reference white color temperature. In the correction method,
The sum obtained by weighted addition of the count number of each gray candidate pixel for each gray candidate image group so that the weight of the count number is higher for pixels closer to the gray black body locus curve, or
The sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of gray pixels in the sample color image to the total number of pixels of the sample color image Correction coefficient according to the presence frequency of the gray candidate pixel group, the gray candidate pixel group for each gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels of the color image. The value obtained by correcting the number,
Is the number of pixels in the gray candidate pixel group. The image signal of each pixel of the color image is multiplied by a predetermined coefficient, and as a result, a skin color candidate pixel group corresponding to each coefficient is obtained by setting a pixel that is in the vicinity of the skin color black body locus curve as a skin color candidate pixel. The coefficient that maximizes the number of pixels of the skin color candidate pixel group is the optimum coefficient, and the skin color candidate pixel group corresponding to the optimum coefficient is the skin color pixel group. A white balance that obtains a color temperature of the light source for photographing the color image from the color temperature and corrects an image signal of the color image multiplied by the optimum coefficient by a difference between the obtained color temperature and a color temperature of a reference white In the correction method,
The sum obtained by weighting and adding the count number of each skin color candidate pixel for each skin color candidate pixel group so that the weight of the count number becomes higher as the pixel closer to the skin color black body locus curve, or
The sample color image for each of the ratios statistically obtained using a large number of the sample color images with respect to the ratio of the number of skin color pixels in the sample color image to the total number of pixels of the sample color image The correction coefficient according to the presence frequency of the skin color candidate pixel group is determined for each skin color candidate pixel group according to the ratio of the number of skin color candidate pixels included in the skin color candidate pixel group to the total number of pixels of the color image. The value obtained by correcting the number,
Is the number of pixels of the skin color candidate pixel group. The gray candidate pixel group corresponding to each coefficient is determined by multiplying the image signal of each pixel of the color image by a predetermined coefficient and, as a result, a pixel that enters the vicinity of the gray black body locus curve is a gray candidate pixel, and skin color A skin color candidate pixel group corresponding to each coefficient is obtained using a pixel that is in the vicinity of the black body locus curve as a skin color candidate pixel, and the sum of the number of pixels of the gray candidate pixel group and the number of pixels of the skin color candidate pixel group And the gray candidate pixel group and the skin color candidate pixel group corresponding to the optimal coefficient as a gray pixel group and a skin color pixel group, respectively, and each pixel included in the gray pixel group The color temperature of the photographing light source of the color image is obtained from the color temperature of each pixel included in the skin color pixel group, and the image signal of the color image multiplied by the optimum coefficient is used as the obtained color. Temperature and In the white balance correction method for correcting only the difference between the quasi-white,
Sum obtained by weighting and adding the count number of each gray candidate pixel for each gray candidate image group so that the weight of the count number becomes higher as the pixel is closer to the gray black body locus curve, or a sample color image The presence frequency of the sample color image for each of the proportions statistically obtained using a number of the sample color images with respect to the proportion of the number of gray pixels in the total number of pixels of the sample color image. The number of gray candidate pixels included in the gray candidate pixel group is corrected for each gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels in the color image. The value obtained as a result, the number of pixels of the gray candidate pixel group, and / or
Sum obtained by weighting and adding the count number of each skin color candidate pixel for each skin color candidate pixel group so that the weight of the count number becomes higher as the pixel closer to the flesh color black body locus curve, or a sample color image The presence frequency of the sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of skin color pixels to the total number of pixels of the sample color image The number of flesh color candidate pixels included in the flesh color candidate pixel group is corrected for each flesh color candidate pixel group for each flesh color candidate pixel group according to the ratio of the total number of pixels in the color image. A white balance correction method comprising a step of setting the obtained value as the number of pixels of the skin color candidate pixel group. The gray candidate pixel group corresponding to each coefficient is determined by multiplying the image signal of each pixel of the color image by a predetermined coefficient and, as a result, a pixel that enters the vicinity of the gray black body locus curve is a gray candidate pixel, and skin color A skin color candidate pixel group corresponding to each coefficient is obtained as a skin color candidate pixel that is in the vicinity of the black body locus curve, and an average color temperature of the gray candidate pixel group and an average color temperature of the skin color candidate pixel group are obtained. The coefficient having the smallest difference is set as an optimum coefficient, and the gray candidate pixel group and the skin color candidate pixel group corresponding to the optimum coefficient are set as a gray pixel group and a skin color pixel group, respectively. The color temperature of the photographing light source of the color image is obtained from the color temperature of the pixel and the color temperature of each pixel included in the skin color pixel group, and the image signal of the color image multiplied by the optimum coefficient is obtained. color In the white balance correction method for correcting by the difference between the degree and reference white,
A value obtained by performing a weighted average operation on the color temperature of each gray candidate pixel included in the gray candidate pixel group so that the gray candidate pixel closer to the gray black body locus curve has a higher color temperature weight. A step of setting an average color temperature of a gray candidate pixel group, and / or
A value obtained by performing a weighted average operation on the color temperature of each skin color candidate pixel included in the skin color candidate pixel group so that the skin temperature candidate pixel closer to the skin color black body locus curve has a higher color temperature weight. A white balance correction method comprising a step of setting an average color temperature of a skin color candidate pixel group. The gray candidate pixel group corresponding to each coefficient is multiplied by a predetermined coefficient for the image signal of each pixel of the color image, and as a result, pixels that enter the vicinity of the gray black body locus curve are gray candidate pixels, and skin color A skin color candidate pixel group corresponding to each coefficient is obtained using a pixel that is in the vicinity of the black body locus curve as a skin color candidate pixel, and the sum of the number of pixels of the gray candidate pixel group and the number of pixels of the skin color candidate pixel group The coefficient that minimizes the difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group is the optimum coefficient, and the gray candidate pixel corresponding to the optimum coefficient The color image is captured from the color temperature of each pixel included in the gray pixel group and the color temperature of each pixel included in the skin color pixel group. Light source Determined temperature, the image signal of the color image the optimum coefficients is multiplied, in the white balance correction method to correct only the difference between the determined color temperature and reference white,
Sum obtained by weighting and adding the count number of each gray candidate pixel for each gray candidate image group so that the weight of the count number becomes higher as the pixel is closer to the gray black body locus curve, or a sample color image The presence frequency of the sample color image for each of the proportions statistically obtained using a number of the sample color images with respect to the proportion of the number of gray pixels in the total number of pixels of the sample color image. The number of gray candidate pixels included in the gray candidate pixel group is corrected for each gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels in the color image. The value obtained as a result, the number of pixels of the gray candidate pixel group, and / or
Sum obtained by weighting and adding the count number of each skin color candidate pixel for each skin color candidate pixel group so that the weight of the count number becomes higher as the pixel closer to the flesh color black body locus curve, or a sample color image The presence frequency of the sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of skin color pixels to the total number of pixels of the sample color image The number of flesh color candidate pixels included in the flesh color candidate pixel group is corrected for each flesh color candidate pixel group for each flesh color candidate pixel group according to the ratio of the total number of pixels in the color image. And / or a step of setting the obtained value as the number of pixels of the skin color candidate pixel group, and / or
A value obtained by performing a weighted average operation on the color temperature of each gray candidate pixel included in the gray candidate pixel group so that the gray candidate pixel closer to the gray black body locus curve has a higher color temperature weight. A step of setting an average color temperature of a gray candidate pixel group, and / or
A value obtained by performing a weighted average operation on the color temperature of each skin color candidate pixel included in the skin color candidate pixel group so that the skin temperature candidate pixel closer to the skin color black body locus curve has a higher color temperature weight. A white balance correction method comprising a step of setting an average color temperature of a skin color candidate pixel group.   6. The white balance correction method according to claim 1, wherein the candidate pixel group is obtained from only pixels effective for obtaining the candidate pixel group among the pixels of the color image.   Average color temperature obtained by weighted average calculation of the color temperature of each pixel included in the candidate pixel group corresponding to the optimum coefficient so that the closer to the black body locus curve, the higher the color temperature weight. The white balance correction method according to claim 1, wherein the color temperature of the photographing light source is obtained.   8. The white balance correction method according to claim 1, wherein a daylight locus curve is used instead of the black body locus curve. Multiplication means for multiplying the image signal of each pixel of the color image by a predetermined coefficient;
As a result of the multiplication, a pixel that falls in the vicinity of a gray black body locus curve is used as a gray candidate pixel to obtain a gray candidate pixel group corresponding to each coefficient, and the number of gray candidate pixel groups for obtaining the number of gray candidate pixel groups Acquisition means;
Optimal coefficient acquisition means for acquiring, as an optimal coefficient, the coefficient corresponding to the gray candidate pixel group having the maximum number of pixels acquired by the gray candidate pixel group pixel number acquisition means;
The gray candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition unit is set as a gray pixel group, and the color temperature of the photographing light source of the color image is determined from the color temperature of each pixel included in the gray pixel group. A light source color temperature acquisition means to be obtained;
The image signal of the color image is multiplied by the optimum coefficient, and the multiplied image signal is corrected by the difference between the color temperature obtained by the light source color temperature acquisition unit and the color temperature of the reference white. And a white balance correction device having correction means for
The gray candidate pixel group pixel number acquisition unit weights and adds the count number of each gray candidate pixel for each gray candidate image group so that the weight closer to the gray black body locus curve is higher in the count number. Or the sum obtained
The sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of gray pixels in the sample color image to the total number of pixels of the sample color image Correction coefficient according to the presence frequency of the gray candidate pixel group, the gray candidate pixel group for each gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels of the color image. The value obtained by correcting the number,
Is the number of pixels of the gray candidate pixel group. Multiplication means for multiplying the image signal of each pixel of the color image by a predetermined coefficient;
As a result of the multiplication, a pixel that enters the vicinity of the skin color black body locus curve is used as a skin color candidate pixel to obtain a skin color candidate pixel group corresponding to each of the coefficients, and a skin color candidate pixel group pixel number for obtaining the number of pixels of the skin color candidate pixel group Acquisition means;
Optimal coefficient acquisition means for acquiring, as an optimal coefficient, the coefficient corresponding to the skin color candidate pixel group in which the number of pixels acquired by the skin color candidate pixel group pixel number acquisition means is maximized;
The skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition unit is defined as a skin color pixel group, and the color temperature of the photographing light source of the color image is determined from the color temperature of each pixel included in the skin color pixel group. A light source color temperature acquisition means to be obtained;
The image signal of the color image is multiplied by the optimum coefficient, and the multiplied image signal is corrected by the difference between the color temperature obtained by the light source color temperature acquisition unit and the color temperature of the reference white. And a white balance correction device having correction means for
The skin color candidate pixel group pixel number acquisition unit weights and adds the count number of each skin color candidate pixel for each skin color candidate pixel group so that a pixel closer to the skin color black body locus curve has a higher weight of the count number. Or the sum obtained
The sample color image for each of the ratios statistically obtained using a large number of the sample color images with respect to the ratio of the number of skin color pixels in the sample color image to the total number of pixels of the sample color image The correction coefficient according to the presence frequency of the skin color candidate pixel group is determined for each skin color candidate pixel group according to the ratio of the number of skin color candidate pixels included in the skin color candidate pixel group to the total number of pixels of the color image. The value obtained by correcting the number,
Is the number of pixels of the skin color candidate pixel group. Multiplication means for multiplying the image signal of each pixel of the color image by a predetermined coefficient;
As a result of the multiplication, pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are defined as skin color candidate pixels. A candidate pixel group pixel number acquisition unit for obtaining a skin color candidate pixel group corresponding to a coefficient and calculating the number of pixels of the gray candidate pixel group and the number of pixels of the skin color candidate pixel group;
Optimal coefficient acquisition means for acquiring, as an optimal coefficient, the coefficient that maximizes the sum of the number of pixels of the gray candidate pixel group acquired by the candidate pixel group pixel number acquisition means and the number of pixels of the skin color candidate pixel group;
The gray candidate pixel group and the skin color candidate pixel group corresponding to the appropriate coefficient acquired by the optimal coefficient acquisition unit are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and Light source color temperature obtaining means for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction unit that multiplies the image signal of the color image by the optimum coefficient and corrects the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition unit and a reference white color. In a white balance correction device having
The candidate pixel group pixel number acquisition unit weights and adds the count number of each gray candidate pixel for each gray candidate image group so that the weight closer to the gray black body locus curve is higher in the count number. Or the ratio of the number of gray pixels in the sample color image to the total number of pixels in the sample color image obtained statistically using a number of sample color images. The gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels of the color image, with a correction coefficient according to the frequency of existence of the sample color image for each ratio. A value obtained by correcting the number of gray candidate pixels for each of the gray candidate pixel groups, and / or
A sum obtained by weighting and adding the count number of each skin color candidate pixel for each skin color candidate pixel group so that the weight of the count number becomes higher as the pixel closer to the skin color black body locus curve, or a sample color image The presence frequency of the sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of skin color pixels to the total number of pixels of the sample color image The number of flesh color candidate pixels included in the flesh color candidate pixel group is corrected for each flesh color candidate pixel group according to the proportion of the total number of pixels in the color image. A white balance correction apparatus that performs a process of setting the obtained value as the number of pixels of the skin color candidate pixel group. Multiplication means for multiplying the image signal of each pixel of the color image by a predetermined coefficient;
As a result of the multiplication, pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are defined as skin color candidate pixels. A candidate color group obtaining means for obtaining a skin color candidate pixel group corresponding to a coefficient, and obtaining an average color temperature of the gray candidate pixel group and an average color temperature of the skin color candidate pixel group;
An optimum coefficient obtaining unit that obtains, as an optimum coefficient, the coefficient that minimizes the difference between the average color temperature of the gray candidate pixel group obtained from the candidate pixel group average color temperature and the average color temperature of the skin color candidate pixel group; ,
The gray candidate pixel group and the skin color candidate pixel group corresponding to the appropriate coefficient acquired by the optimal coefficient acquisition unit are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and Light source color temperature obtaining means for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction unit that multiplies the image signal of the color image by the optimum coefficient and corrects the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition unit and a reference white color. In a white balance correction device having
The candidate pixel group average color temperature acquisition unit is configured so that a gray candidate pixel closer to the gray black body locus curve has a higher color temperature weight so that the color temperature of each gray candidate pixel included in the gray candidate pixel group is higher. A value obtained by performing a weighted average calculation on the gray candidate pixel group as an average color temperature, and / or
A value obtained by performing a weighted average operation on the color temperature of each skin color candidate pixel included in the skin color candidate pixel group so that the skin temperature candidate pixel closer to the skin color black body locus curve has a higher color temperature weight. A white balance correction apparatus that performs processing to obtain an average color temperature of a skin color candidate pixel group. Multiplication means for multiplying the image signal of each pixel of the color image by a predetermined coefficient;
As a result of the multiplication, pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are each defined as the skin color candidate pixel. Candidate pixel group obtaining means for obtaining a skin color candidate pixel group corresponding to the coefficient;
Candidate pixel group pixel number obtaining means for obtaining the number of pixels of the gray candidate pixel group and the number of pixels of the skin color candidate pixel group obtained by the candidate pixel group obtaining means;
Candidate pixel group average color temperature acquisition means for obtaining an average color temperature of the gray candidate pixel group acquired by the candidate pixel group acquisition means and an average color temperature of the skin color candidate pixel group;
The sum of the number of pixels of the gray candidate pixel group acquired by the candidate pixel group pixel number acquisition unit and the number of pixels of the skin color candidate pixel group is maximized, and acquired by the candidate pixel group average color temperature acquisition unit An optimum coefficient obtaining unit that obtains, as an optimum coefficient, the coefficient that minimizes the difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group;
The gray candidate pixel group and the skin color candidate pixel group corresponding to the appropriate coefficient acquired by the optimal coefficient acquisition unit are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and Light source color temperature obtaining means for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction unit that multiplies the image signal of the color image by the optimum coefficient and corrects the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition unit and a reference white color. In a white balance correction device having
The candidate pixel group pixel number acquisition unit weights and adds the count number of each gray candidate pixel for each gray candidate image group so that the weight closer to the gray black body locus curve is higher in the count number. Or the ratio of the number of gray pixels in the sample color image to the total number of pixels in the sample color image obtained statistically using a number of sample color images. The gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels of the color image, with a correction coefficient according to the frequency of existence of the sample color image for each ratio. A value obtained by correcting the number of gray candidate pixels for each of the gray candidate pixel groups, and / or
A sum obtained by weighting and adding the count number of each skin color candidate pixel for each skin color candidate pixel group so that the weight of the count number becomes higher as the pixel closer to the skin color black body locus curve, or a sample color image The presence frequency of the sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of skin color pixels to the total number of pixels of the sample color image The number of flesh color candidate pixels included in the flesh color candidate pixel group is corrected for each flesh color candidate pixel group according to the proportion of the total number of pixels in the color image. A white balance correction apparatus that performs a process of setting the obtained value as the number of pixels of the skin color candidate pixel group. Multiplication means for multiplying the image signal of each pixel of the color image by a predetermined coefficient;
As a result of the multiplication, pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are each defined as the skin color candidate pixel. Candidate pixel group obtaining means for obtaining a skin color candidate pixel group corresponding to the coefficient;
Candidate pixel group pixel number obtaining means for obtaining the number of pixels of the gray candidate pixel group and the number of pixels of the skin color candidate pixel group obtained by the candidate pixel group obtaining means;
Candidate pixel group average color temperature acquisition means for obtaining an average color temperature of the gray candidate pixel group acquired by the candidate pixel group acquisition means and an average color temperature of the skin color candidate pixel group;
The sum of the number of pixels of the gray candidate pixel group acquired by the candidate pixel group pixel number acquisition unit and the number of pixels of the skin color candidate pixel group is maximized, and acquired by the candidate pixel group average color temperature acquisition unit An optimum coefficient obtaining unit that obtains, as an optimum coefficient, the coefficient that minimizes the difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group;
The gray candidate pixel group and the skin color candidate pixel group corresponding to the appropriate coefficient acquired by the optimal coefficient acquisition unit are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and Light source color temperature obtaining means for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction unit that multiplies the image signal of the color image by the optimum coefficient and corrects the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition unit and a reference white color. In a white balance correction device having
The candidate pixel group average color temperature acquisition unit is configured so that a gray candidate pixel closer to the gray black body locus curve has a higher color temperature weight so that the color temperature of each gray candidate pixel included in the gray candidate pixel group is higher. A value obtained by performing a weighted average calculation on the gray candidate pixel group as an average color temperature, and / or
A value obtained by performing a weighted average operation on the color temperature of each skin color candidate pixel included in the skin color candidate pixel group so that the skin temperature candidate pixel closer to the skin color black body locus curve has a higher color temperature weight. A white balance correction apparatus that performs processing to obtain an average color temperature of a skin color candidate pixel group.   Of the pixels of the color image, effective pixels are extracted to obtain the gray candidate pixel group and / or skin color candidate pixel group, and only the effective pixels are extracted from the gray candidate pixel group and / or skin color candidate pixel group. The white balance correction apparatus according to claim 9, further comprising effective pixel extraction means for obtaining the image data.   The light source color temperature acquisition unit calculates a weighted average of the color temperatures of each pixel included in the candidate pixel group corresponding to the optimal coefficient so that a pixel closer to the black body locus curve has a higher color temperature weight. 16. The white balance correction apparatus according to claim 9, wherein an average color temperature obtained by calculation is obtained as a color temperature of the photographing light source.   The white balance correction device according to any one of claims 9 to 16, wherein a daylight locus curve is used instead of the black body locus curve. A multiplication process for multiplying an image signal of each pixel of the color image by a predetermined coefficient;
As a result of the multiplication, a pixel that enters the vicinity of the gray black body locus curve is regarded as a gray candidate pixel, a gray candidate pixel group corresponding to each coefficient is obtained, and the number of gray candidate pixel groups for obtaining the number of pixels in the gray candidate pixel group Acquisition process,
An optimal coefficient acquisition process for acquiring, as an optimal coefficient, the coefficient corresponding to the gray candidate pixel group in which the number of pixels acquired by the gray candidate pixel group pixel number acquisition process is maximized;
The gray candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process is defined as a gray pixel group, and the color temperature of the imaging light source of the color image is determined from the color temperature of each pixel included in the gray pixel group. The desired light source color temperature acquisition process;
The image signal of the color image is multiplied by the optimum coefficient, and the multiplied image signal is corrected by the difference between the color temperature obtained by the light source color temperature acquisition process and the color temperature of the reference white. A program for causing a computer to perform correction processing,
In the gray candidate pixel group pixel number acquisition process, the count number of each gray candidate pixel is weighted and added for each gray candidate image group so that the weight closer to the gray black body locus curve is higher in the count number. Or the sum obtained
The sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of gray pixels in the sample color image to the total number of pixels of the sample color image Correction coefficient according to the presence frequency of the gray candidate pixel group, the gray candidate pixel group for each gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels of the color image. The value obtained by correcting the number,
Is the number of pixels of the gray candidate pixel group. A multiplication process for multiplying an image signal of each pixel of the color image by a predetermined coefficient;
As a result of the multiplication, a pixel that enters the vicinity of the skin color black body locus curve is used as a skin color candidate pixel to obtain a skin color candidate pixel group corresponding to each of the coefficients, and a skin color candidate pixel group pixel number for obtaining the number of pixels of the skin color candidate pixel group Acquisition process,
An optimal coefficient acquisition process for acquiring the coefficient corresponding to the skin color candidate pixel group that maximizes the number of pixels acquired by the skin color candidate pixel group pixel number acquisition process;
The skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process is defined as a skin color pixel group, and the color temperature of the photographing light source of the color image is determined from the color temperature of each pixel included in the skin color pixel group. The desired light source color temperature acquisition process;
The image signal of the color image is multiplied by the optimum coefficient, and the multiplied image signal is corrected by the difference between the color temperature obtained by the light source color temperature acquisition process and the color temperature of the reference white. A program for causing a computer to execute correction processing to be performed,
In the skin color candidate pixel group pixel number acquisition process, the count number of each skin color candidate pixel is weighted and added for each skin color candidate pixel group so that the weight closer to the skin color black body locus curve is higher in the count number. Or the sum obtained
The sample color image for each of the ratios statistically obtained using a large number of the sample color images with respect to the ratio of the number of skin color pixels in the sample color image to the total number of pixels of the sample color image The correction coefficient according to the presence frequency of the skin color candidate pixel group is determined for each skin color candidate pixel group according to the ratio of the number of skin color candidate pixels included in the skin color candidate pixel group to the total number of pixels of the color image. The value obtained by correcting the number,
Is a number of pixels of the corresponding skin color candidate pixel group. A multiplication process for multiplying the image signal of each pixel of the color image by a predetermined coefficient;
As a result of the multiplication, pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are defined as skin color candidate pixels. Obtaining a skin color candidate pixel group corresponding to a coefficient, obtaining a pixel number of the gray candidate pixel group and a number of pixels of the skin color candidate pixel group, and obtaining a pixel number of candidate pixel groups,
An optimum coefficient obtaining process for obtaining, as an optimum coefficient, the coefficient that maximizes the sum of the number of pixels of the gray candidate pixel group obtained by the candidate pixel group pixel number obtaining process and the number of pixels of the skin color candidate pixel group;
The gray candidate pixel group and the skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and A light source color temperature acquisition process for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction process for multiplying the image signal of the color image by the optimum coefficient, and correcting the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition process and a reference white color Is a program that causes a computer to execute
In the candidate pixel group pixel number acquisition processing, the count number of each gray candidate pixel is weighted and added for each gray candidate image group so that the weight closer to the gray black body locus curve is higher in the count number. Or the ratio of the number of gray pixels in the sample color image to the total number of pixels in the sample color image, statistically determined using a number of sample color images. The gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels of the color image with a correction coefficient according to the frequency of existence of the sample color image for each ratio. A procedure in which the value obtained by correcting the number of gray candidate pixels for each time is the number of pixels of the gray candidate pixel group, and / or
Sum obtained by weighting and adding the count number of each skin color candidate pixel for each skin color candidate pixel group so that the weight of the count number becomes higher as the pixel closer to the flesh color black body locus curve, or a sample color image The presence frequency of the sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of skin color pixels to the total number of pixels of the sample color image The number of flesh color candidate pixels included in the flesh color candidate pixel group is corrected for each flesh color candidate pixel group for each flesh color candidate pixel group according to the ratio of the total number of pixels in the color image. A program having a procedure of setting the obtained value as the number of pixels of the skin color candidate pixel group. A multiplication process for multiplying the image signal of each pixel of the color image by a predetermined coefficient;
As a result of the multiplication, pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are defined as skin color candidate pixels. Obtaining a skin color candidate pixel group corresponding to a coefficient, and obtaining an average color temperature of the gray candidate pixel group and an average color temperature of the skin color candidate pixel group;
An optimal coefficient acquisition process for acquiring, as an optimal coefficient, the coefficient that minimizes the difference between the average color temperature of the gray candidate pixel group acquired by the candidate pixel group average color temperature and the average color temperature of the skin color candidate pixel group; ,
The gray candidate pixel group and the skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and A light source color temperature acquisition process for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction process for multiplying the image signal of the color image by the optimum coefficient, and correcting the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition process and a reference white color Is a program that causes a computer to execute
In the candidate pixel group average color temperature acquisition process, the color temperature of each gray candidate pixel included in the gray candidate pixel group is such that a gray candidate pixel closer to the gray black body locus curve has a higher color temperature weight. The procedure of setting the value obtained by performing the weighted average calculation as the average color temperature of the gray candidate pixel group, and / or
A value obtained by performing a weighted average operation on the color temperature of each skin color candidate pixel included in the skin color candidate pixel group so that the skin temperature candidate pixel closer to the skin color black body locus curve has a higher color temperature weight. A program having a procedure for setting an average color temperature of a skin color candidate pixel group. A multiplication process for multiplying the image signal of each pixel of the color image by a predetermined coefficient;
As a result of the multiplication, the pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are each defined as the skin color candidate pixel. A candidate pixel group acquisition process for obtaining a skin color candidate pixel group corresponding to the coefficient;
A candidate pixel group pixel number acquisition process for obtaining the number of pixels of the gray candidate pixel group acquired by the candidate pixel group acquisition process and the number of pixels of the skin color candidate pixel group;
A candidate pixel group average color temperature acquisition process for obtaining an average color temperature of the gray candidate pixel group acquired by the candidate pixel group acquisition process and an average color temperature of the skin color candidate pixel group;
The sum of the number of pixels of the gray candidate pixel group acquired by the candidate pixel group pixel number acquisition process and the number of pixels of the skin color candidate pixel group is maximized, and acquired by the candidate pixel group average color temperature acquisition process An optimal coefficient acquisition process for acquiring, as an optimal coefficient, the coefficient that minimizes the difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group;
The gray candidate pixel group and the skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and A light source color temperature acquisition process for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction process for multiplying the image signal of the color image by the optimum coefficient, and correcting the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition process and a reference white color Is a program that causes a computer to execute
The candidate pixel group pixel number acquisition processing weights and adds the count number of each gray candidate pixel for each gray candidate image group so that the weight closer to the gray black body locus curve is higher in the count number. Or the ratio of the number of gray pixels in the sample color image to the total number of pixels in the sample color image obtained statistically using a number of sample color images. The gray candidate pixel group according to the ratio of the number of gray candidate pixels included in the gray candidate pixel group to the total number of pixels of the color image, with a correction coefficient according to the frequency of existence of the sample color image for each ratio. A value obtained by correcting the number of gray candidate pixels for each step, and / or the number of pixels in the gray candidate pixel group, and / or
Sum obtained by weighting and adding the count number of each skin color candidate pixel for each skin color candidate pixel group so that the weight of the count number becomes higher as the pixel closer to the flesh color black body locus curve, or a sample color image The presence frequency of the sample color image for each of the ratios statistically obtained using a number of the sample color images with respect to the ratio of the number of skin color pixels to the total number of pixels of the sample color image The number of flesh color candidate pixels included in the flesh color candidate pixel group is corrected for each flesh color candidate pixel group for each flesh color candidate pixel group according to the ratio of the total number of pixels in the color image. A program having a procedure of setting the obtained value as the number of pixels of the skin color candidate pixel group. A multiplication process for multiplying the image signal of each pixel of the color image by a predetermined coefficient;
As a result of the multiplication, the pixels that are in the vicinity of the gray black body locus curve are regarded as gray candidate pixels, and the gray candidate pixel group corresponding to each coefficient and the pixels that are in the vicinity of the skin color black body locus curve are each defined as the skin color candidate pixel. A candidate pixel group acquisition process for obtaining a skin color candidate pixel group corresponding to the coefficient;
A candidate pixel group pixel number acquisition process for obtaining the number of pixels of the gray candidate pixel group acquired by the candidate pixel group acquisition process and the number of pixels of the skin color candidate pixel group;
A candidate pixel group average color temperature acquisition process for obtaining an average color temperature of the gray candidate pixel group acquired by the candidate pixel group acquisition process and an average color temperature of the skin color candidate pixel group;
The sum of the number of pixels of the gray candidate pixel group acquired by the candidate pixel group pixel number acquisition process and the number of pixels of the skin color candidate pixel group is maximized, and acquired by the candidate pixel group average color temperature acquisition process An optimal coefficient acquisition process for acquiring, as an optimal coefficient, the coefficient that minimizes the difference between the average color temperature of the gray candidate pixel group and the average color temperature of the skin color candidate pixel group;
The gray candidate pixel group and the skin color candidate pixel group corresponding to the optimal coefficient acquired by the optimal coefficient acquisition process are set as a gray pixel group and a skin color pixel group, respectively, and the color temperature of each pixel included in the gray pixel group and A light source color temperature acquisition process for obtaining a color temperature of a photographing light source of the color image from a color temperature of each pixel included in the skin color pixel group;
A correction process for multiplying the image signal of the color image by the optimum coefficient, and correcting the multiplied image signal by a difference between the color temperature obtained by the light source color temperature acquisition process and a reference white color Is a program that causes a computer to execute
In the candidate pixel group average color temperature acquisition process, the color temperature of each gray candidate pixel included in the gray candidate pixel group is such that a gray candidate pixel closer to the gray black body locus curve has a higher color temperature weight. The procedure of setting the value obtained by performing the weighted average calculation as the average color temperature of the gray candidate pixel group, and / or
A value obtained by performing a weighted average operation on the color temperature of each skin color candidate pixel included in the skin color candidate pixel group so that the skin temperature candidate pixel closer to the skin color black body locus curve has a higher color temperature weight. A program having a procedure for setting an average color temperature of a skin color candidate pixel group.   Of the pixels of the color image, effective pixels are extracted to obtain the gray candidate pixel group and / or skin color candidate pixel group, and only the effective pixels are extracted from the gray candidate pixel group and / or skin color candidate pixel group. The program according to claim 9, further causing a computer to perform an effective pixel extraction process for obtaining the data.   In the light source color temperature acquisition process, a weighted average of the color temperatures of each pixel included in the candidate pixel group corresponding to the optimal coefficient is set such that a pixel closer to the black body locus curve has a higher color temperature weight. 25. The program according to claim 18, wherein an average color temperature obtained by calculation is obtained as a color temperature of the photographing light source.   The program according to any one of claims 18 to 25, wherein a daylight locus curve is used instead of the black body locus curve.

Images