Color-balance correction adapted to tendency of color fogging

The present invention relates to a color balance correction of an image, and more particularly to a technique of performing a preferable color balance correction according to a tendency of color cast generated in an image.

For example, when an image is generated by imaging with a digital still camera, a color cast may occur in which the influence of a specific color corresponding to the light source appears on the entire image due to the influence of the light source at the time of imaging. Generally, color balance correction is performed in order to correct the color cast of an image.

Regarding the color balance correction of an image, a technique for correcting the color so that the human skin color is naturally expressed and the entire image has a good color balance is disclosed (for example, Patent Document). 1). This technology does not perform correction using a correction amount that makes the human skin color expressed in the image a preferable ideal color for human skin color, but is constant in the difference between the skin color expressed in the image and the ideal color. The entire image is corrected using the correction amount calculated by multiplying the reduction coefficient of. In such correction, the skin color expressed in the image is corrected so as to approach the ideal color, and the reduced correction amount is used, so that the color balance of the entire image is well maintained.

<patcit num="1"><text>JP 2002-44469</text></patcit><patcit num="2"><text>JP 2001-320727</text></patcit>

<p> The color cast that occurs in an image has a tendency such as the color that has an effect and the degree of the effect that appears, depending on the imaging conditions such as the type of light source. The color cast of the image may give different impressions to the observer due to this difference in tendency. For example, the color cast in the green direction, which often occurs when an image is taken under a fluorescent lamp, may give an unnatural impression to the observer. On the other hand, the color cast in the orange direction, which often occurs when an image is taken under an incandescent lamp, may not give the observer an unnatural impression. Therefore, the above-mentioned reduction coefficient when performing color balance correction, that is, the ratio of the correction amount to the difference between the skin color expressed in the image and the ideal color, may differ from the preferable value depending on the tendency of color cast.</p><p> However, in the above-mentioned conventional technique, the ratio of the correction amount to the difference between the skin color expressed in the image and the ideal color is constant regardless of the difference in the tendency of color cast. Therefore, there is a problem that it is difficult to perform a preferable color balance correction according to the tendency of color cast.</p><p> It should be noted that such a problem is not limited to the image generated by the digital still camera, but is a common problem when performing color balance correction of the image.</p><p> The present invention has been made to solve the above-mentioned conventional problems, and an object of the present invention is to provide a technique capable of performing preferable color balance correction according to a tendency of color cast.</p>

<p> In order to solve at least a part of the above problems, the image processing apparatus of the present invention is used. An image processing device that corrects the color balance of an image composed of a plurality of pixels. A representative value calculation unit that calculates representative values representing the hue and saturation of the pixels included in the skin color region based on the signal values of the pixels included in the skin color region including the pixels representing human skin in the image. The image using an execution correction amount set based on a standard correction amount obtained by multiplying the difference between the representative value and the ideal value set as a value representing the ideal hue and saturation of human skin by a reduction coefficient. It is equipped with a correction processing unit that corrects the color balance of the image. The value of the reduction coefficient is variably set according to the representative value.</p><p> In this image processing device, the difference between a representative value representing the hue and saturation of human skin expressed in an image and an ideal value set as a value representing the ideal hue and saturation of human skin. The value of the reduction coefficient, which indicates the ratio of the standard correction amount to the standard value, is variably set according to the representative value. Further, this image processing device performs correction using an execution correction amount set based on the standard correction amount. Therefore, in this image processing apparatus, it is possible to perform preferable color balance correction according to the tendency of color cast.</p><p> In the image processing apparatus, the representative value and the ideal value are values representing positions on a plane representing hue and saturation, and are values. The difference between the representative value and the ideal value may be a value representing the distance between the representative value and the ideal value on the plane.</p><p> Further, in the image processing apparatus, the reduction coefficient is a value when the representative value is located at the first point on the first straight line from the position of the ideal value on the plane to the first direction. The representative value is a point on the second straight line from the position of the ideal value on the plane toward the second direction, and the distance from the position of the ideal value is the position of the ideal value and the first position. The values when located at a second point equal to the distance between the points may be set differently from each other.</p><p> In this way, even if the position of the representative value is the same as the distance from the position of the ideal value, if it is located on a straight line heading in a different direction, the value of the reduction coefficient can be made different. Therefore, it is possible to perform preferable color balance correction according to the direction of color cast.</p><p> Further, in the image processing apparatus, the reduction coefficient is a value when the representative value is located at a first point on a straight line passing through the position of the ideal value on the plane, and the representative value is on the plane. The value when the value is located at a second point different from the first point on the straight line may be set so as to be different from each other.</p><p> In this way, even when the representative value is located on one straight line passing through the position of the ideal value, the value of the reduction coefficient can be changed according to the distance from the position of the ideal value. Therefore, it is possible to perform preferable color balance correction according to the degree of color cast.</p><p> Further, in the image processing apparatus, the correction processing unit may calculate the value of the reduction coefficient according to the representative value by using the reference reduction coefficient set for the plurality of positions on the plane.</p><p> By doing so, the reduction coefficient can be easily calculated, and the processing speed can be increased.</p><p> Further, in the image processing apparatus, the reference reduction coefficient is set along a plurality of reference straight lines connecting the position of the ideal value on the plane and each of the plurality of predetermined reference points. The correction processing unit is two points on the reference straight line that are closest to the position of the representative value on the plane, and the distance from the position of the ideal value is the position of the ideal value and the position of the ideal value. The reduction coefficient is calculated using the reference reduction coefficient at a point that is the same as the distance between the positions of the representative values and the distance between the positions of the representative values and the reference points on the two reference lines. You may do so.</p><p> In this way, the tendency of the reduction coefficient according to the tendency of the color cast can be easily set, and the setting for performing the preferable color balance correction can be easily performed.</p><p> The present invention can be realized in various aspects, for example, for realizing an image processing method and device, an image correction method and device, an image conversion method and device, and a function of these methods or devices. It can be realized in the form of a computer program, a recording medium on which the computer program is recorded, a data signal including the computer program and embodied in a carrier, and the like.</p>

Next, embodiments of the present invention will be described in the following order based on examples. A. Example: B. Modification example:

A. Example: FIG. 1 is an explanatory diagram schematically showing a configuration of an image processing apparatus as an embodiment of the present invention. The image processing device 100 of the embodiment is configured as a computer, and includes a CPU 110, a display unit 120 such as a monitor, an operation unit 130 such as a keyboard and a mouse, an external storage device 140 such as a hard disk drive, and an interface unit ( It is equipped with an I / F section) 150 and an internal storage device 200 such as ROM and RAM. Each component of the image processing apparatus 100 is connected to each other via a bus 160.

The interface unit 150 is provided with a plurality of input / output terminals, and exchanges information with external devices. For example, the interface unit 150 is connected to the digital still camera 300 via a cable and reads the image data recorded in the digital still camera 300. Further, the interface unit 150 is connected to the printer 400 via a cable, and supplies print data for printing an image to the printer 400.

The color balance correction unit 210 is stored in the internal storage device 200. The color balance correction unit 210 is a computer program for executing color balance correction processing of an image under a predetermined operating system. The CPU 110 executes the color balance correction process by reading and executing this program from the internal storage device 200.

The color balance correction unit 210 includes a face area input unit 212, a representative value calculation unit 214, and a correction processing unit 216 as modules. Further, the correction processing unit 216 includes a reduction coefficient calculation unit 217 and a correction amount calculation unit 218 as modules. Further, the face area input unit 212 may include the face area detection unit 213 as a module. The functions of each of these parts will be described in detail in the description of the color balance correction process described later.

FIG. 2 is a flowchart showing the flow of color balance correction processing by the image processing apparatus 100 of the embodiment. The color balance correction process of this embodiment is a process of calculating a correction amount according to a tendency of color cast expected from the human skin color expressed in the image and performing color balance correction using the correction amount. ..

In step S110, the color balance correction unit 210 (FIG. 1) inputs the target image data to be the target of the color balance correction. The target image data may be input from the digital still camera 300 (FIG. 1) or from the external storage device 140 (FIG. 1). Further, the target image data may be input via an externally connected device such as a CD-R / RW drive (not shown) or the Internet. The input target image data is stored in a predetermined area in the internal storage device 200. In this embodiment, the target image data is input as RGB data, but the target image data may be input as data expressed in another color space.

In step S120 (FIG. 2), the face area input unit 212 (FIG. 1) inputs information for identifying the face area in the target image represented by the input target image data. Here, the face area is an area expressing a human face in the target image. Further, the information for specifying the face region is information for identifying pixels included in the face region of the target image, and is, for example, information indicating the position of the pixel. The information for identifying the face area may be input, for example, by the user performing an operation of designating the face area on the target image displayed on the display unit 120 (FIG. 1). Further, the face area detection unit 213 (FIG. 1) included in the face area input unit 212 identifies the face area by using a known face area detection method (for example, the method described in Japanese Patent Application Laid-Open No. 2001-16573). You may enter the information.

In step S130 (FIG. 2), the representative value calculation unit 214 (FIG. 1) identifies the face region from the target image and calculates the skin color representative value FV from the image data corresponding to the face region. Here, the skin color representative value FV is a value representing the human skin color expressed in the target image. In this embodiment, as the skin color representative value FV, the RGB value calculated by averaging the pixel values (RGB values) of all the pixels included in the face area of the target image for each of the R value, G value, and B value is adopted. ing. In addition, various other indexes can be adopted as the skin color representative value FV, and the skin color representative value FV is the R value, G value, and B value of the pixel values of all the pixels included in the face area of the target image. It may be the peak value in each histogram or the median value in a similar histogram.

In step S140 (Fig. 2), the representative value calculation unit 214 (Fig. 1) sets the skin color representative value FV calculated as an RGB value to L.^*a^*b b^*Convert to a value and calculate the converted skin color representative value FVc. This L^*a^*b b^*After conversion as a value Skin color representative value FVc a^*Value and b^*The value is a value representing the hue and saturation of the human skin color expressed in the target image. The converted skin color representative value FVc is used as an index showing the tendency of the color cast of the target image in relation to the ideal value of the skin color described later.

In step S150 (FIG. 2), the reduction coefficient calculation unit 217 (FIG. 1) of the correction processing unit 216 calculates the reduction coefficient Kf based on the converted skin color representative value FVc. As will be described later, the reduction coefficient Kf represents the ratio of the correction amount to the difference between the human skin color expressed in the target image and the ideal skin color, and takes various values in the range of 0 to 0.8. Set. The calculation method of the reduction coefficient Kf will be described in detail later.

In step S160 (FIG. 2), the correction amount calculation unit 218 (FIG. 1) of the correction processing unit 216 calculates the standard correction amount Δ (N). FIG. 3 is an explanatory diagram showing an outline of a method for calculating the standard correction amount Δ (N). Figure 3 shows a as a plane representing hue and saturation.^*-b^*The skin color representative point FP and the ideal point IP plotted on a plane are shown. Here, the skin color representative point FP is a of the above-mentioned converted skin color representative value FVc.^*-b^*It is a point that represents a position on a plane. The ideal point IP is a, which is an ideal value representing the ideal hue and saturation of human skin color.^*-b^*It is a point that represents a position on a plane. In this embodiment, the standard correction amount Δ (N) is a.^*-b^*A on the plane^*Direction standard correction amount Δa^*(N) and b^*Direction standard correction amount Δb^*Expressed as a combination with (N). As shown in FIG. 3, the standard correction amount Δ (N) (Δa)^*(N) and Δb^*(N)) is the value of the skin color representative point FP (a of the converted skin color representative value FVc).^*Value and b^*Value) and ideal point IP value (a)^*Value and b^*Difference from (value) ("Δa"^*₀And "Δb^*₀It is calculated by multiplying each of) by the reduction coefficient Kf. That is, as described above, the reduction coefficient Kf is the difference (Δa) between the value of the skin color representative point FP and the value of the ideal point IP.^*₀And Δb^*₀) With respect to the standard correction amount Δ (N) (Δa)^*(N) and Δb^*It is used as a ratio of (N)). The difference between the value of the skin color representative point FP and the value of the ideal point IP (Δa^*₀And "Δb^*₀) Is a^*-b^*It is the distance between the skin color representative point FP and the ideal point IP on a plane.

The position of the ideal point IP is set by reflecting the result of investigating the user's preference. The position of this ideal point IP may be changeable by the user.

The standard correction amount Δ (N) calculated in this way is that the pixel having the value of the skin color representative point FP is a point on the line segment connecting the skin color representative point FP and the ideal point IP, and the skin color representative point FP. The correction amount is such that the distance from is corrected to the pixel having the value of the point AP in which the reduction coefficient Kf of the length of the line segment is multiplied. That is, the larger the value of the reduction coefficient Kf, the larger the standard correction amount Δ (N) is so that the pixel having the value of the skin color representative point FP becomes the pixel having the value closer to the value of the ideal point IP. ..

In step S170 (FIG. 2), the correction amount calculation unit 218 (FIG. 1) sets the execution correction amount Δ (E) based on the standard correction amount Δ (N). The execution correction amount Δ (E) is a correction amount actually used when performing color balance correction for all pixels of the target image. FIG. 4 is an explanatory diagram showing an example of a method of setting the execution correction amount Δ (E). In Fig. 4, a of the execution correction amount Δ (E) is shown.^*Direction execution correction amount Δa^*Shows the setting method of (E). As shown in Fig. 4 (a), the execution correction amount Δa^*(E), a before correction^*Standard correction amount Δa uniformly regardless of the value^*It may be set to the same value as (N). In addition, as shown in Fig. 4 (b), a of the skin color representative point FP^*Execution correction amount Δa corresponding to the value^*(E) is the standard correction amount Δa^*Same value as (N), other a^*Execution correction amount Δa corresponding to the value^*(E), a^*It may be set by interpolation of a curve (for example, a spline curve) using a point that takes the maximum value and a point that takes the minimum value. b b^*Direction execution correction amount Δb^*Set (E) in the same way. The execution correction amount Δ (E) may be set based on the standard correction amount Δ (N), and various other setting methods can be adopted.

In step S180 (FIG. 2), the correction processing unit 216 (FIG. 1) corrects the color balance of the target image using the execution correction amount Δ (E). Specifically, for all the pixels that make up the target image, from the RGB value to L^*a^*b b^*Conversion to value and L after conversion^*a^*b b^*Of the value a^*Value and b^*Execution correction amount Δ (E) (Δa) for the value^*(E) and Δb^*Correction using (E)) and L after correction^*a^*b b^*The conversion from the value to the RGB value is executed, and the corrected image data is calculated. For example, when the execution correction amount Δ (E) is set by the method shown in FIG. 4 (a), a of all the pixels constituting the target image^*-b^*Color balance correction is performed so that the position on the plane moves in parallel by the standard correction amount Δ (N). When the execution correction amount Δ (E) is set by the method shown in FIG. 4 (b), a of the skin color representative point FP among the pixels in the target image^*Value (or b^*Pixels with the same value as (value) are a^*-b^*The position on the plane is translated by the standard correction amount Δ (N), and the other pixels are translated by the set execution correction amount Δ (E).

Through the above processing, the target is targeted using the execution correction amount Δ (E) set based on the standard correction amount Δ (N) obtained by multiplying the difference between the skin color representative point FP value and the ideal point IP value by the reduction coefficient Kf. Image color balance correction is performed.

Next, a method of calculating the reduction coefficient Kf will be described. In this embodiment, a^*-b^*The reduction coefficient Kf is calculated using the reference reduction coefficient Ks set along a plurality of reference straight lines SL on the plane. Figure 5 shows a^*-b^*It is explanatory drawing which shows an example of the reference straight line SL on a plane. Here, the reference straight line SL is a line indicating the direction of the reference color cast, and is a straight line connecting the ideal point IP and the reference color cast point SP. Further, the reference color cast point SP is assumed to be a pixel representing a human skin color in an image when a predetermined color cast occurs in an image captured under a predetermined reference light source a.^*-b^*It is a point on a plane. In this embodiment, as shown in FIG. 5, six reference color cast points SP corresponding to the six reference light sources are set. Specifically, the yellow reference color cover point SPy corresponding to the yellow reference light source, the orange reference color cover point SPo corresponding to the orange reference light source, and the red reference color cover point corresponding to the red reference light source. SPr, magenta reference color cover point SPm corresponding to magenta reference light source, blue reference color cover point SPb corresponding to blue reference light source, and green reference color cover point SPg corresponding to green reference light source. , Is set. Six reference straight lines SL (SLy, SLo, SLr, SLm, SLb, SLg) are set corresponding to these six reference color cast points SP.

In addition, a^*-b^*On the plane, as shown in Fig. 5, the human skin color represented in the image is a.^*-b^*A target region TA (shown with hatching in FIG. 5), which is a region preferably located on a plane, is set.

The reference reduction coefficient Ks is set along each reference straight line SL. FIG. 6 is an explanatory diagram showing an example of the reference reduction coefficient Ks. The reference reduction coefficient Ks is a reference reduction coefficient set according to the distance from the ideal point IP indicating the degree of color cast along the reference straight line SL indicating the direction of color cast corresponding to the reference light source. In this embodiment, six reference reduction coefficients Ks are set along the six reference straight lines SL. Fig. 6 (a) shows an example of the reference reduction coefficient Kso along the orange reference line SLo (Fig. 5), and Fig. 6 (b) shows an example along the green reference line SLg (Fig. 5). An example of the reference reduction coefficient Ksg is shown. In FIG. 6, Lt indicates the distance from the ideal point IP.

For example, the value of the reference reduction coefficient Kso along the orange reference line SLo shown in Fig. 6 (a) increases as the distance Lt increases from 0, and the point CPo on the outer circumference of the target region TA (see Fig. 5). It is the maximum value (0.8) in. Furthermore, as the distance Lt increases, the value of the reference reduction coefficient Kso remains at the maximum value for a while and then decreases. At the limit point EPo (see Fig. 5) where the distance Lt is larger than the reference fog point SPo, the value of the reference reduction coefficient Kso is 0.

In addition, the value of the reference reduction coefficient Ksg along the green reference straight line SLg shown in Fig. 6 (b) increases as the distance Lt increases from 0, and the point CPg on the outer circumference of the target area TA (see Fig. 5). It is the maximum value (0.8) in. After this point, the value of the reference reduction coefficient Kso decreases as the distance Lt increases. At the limit point EPg (see Fig. 5) where the distance Lt is larger than the reference fog point SPg, the value of the reference reduction coefficient Ksg is 0.

As described above, in this embodiment, the reference reduction coefficient Ks is set so that the value differs depending on the distance Lt from the ideal point IP even on one reference straight line SL. That is, even if the color cast is in the same direction, the value of the reference reduction coefficient Ks is set to be different depending on the degree of the color cast.

Further, since each reference reduction coefficient Ks is set independently, the tendency of the reference reduction coefficient Ks can be different for each reference straight line SL as shown in FIG. Specifically, within the target area TA (Fig. 5), if the distance Lt from the ideal point IP is the same, the value of the reference reduction coefficient Kso along the orange reference line SLo becomes the green reference line SLg. It is set to be smaller than the value of the reference reduction coefficient Ksg along the line. This is because the color cast in the direction along the reference straight line SLo of orange tends not to require a strong correction as compared with the color cast in the direction along the reference straight line SLg of green. As described above, in this embodiment, the value of the reference reduction coefficient Ks is set to be different depending on the direction of the color cast even if the color cast is the same.

In addition, the distance from the ideal point IP to the limit point EP is set larger for the limit point EPo on the orange reference line SLo than for the limit point EPg on the green reference line SLg. This is because there is a possibility of false detection that pixels that are not human skin color are detected as the face area, but the direction from the ideal point IP along the orange reference straight line SLo is better than the direction along the green reference straight line SLg. Is high.

In this embodiment, the reference reduction coefficient Ks for red and magenta is set to have the same tendency as the reference reduction coefficient Kso for orange. On the other hand, the reference reduction coefficient Ks for blue and yellow is set to have the same tendency as the reference reduction coefficient Ksg for green.

The reference reduction coefficient Ks can be set to have various tendencies according to the user's preference. It is also possible to set a plurality of reference reduction coefficients Ks having different tendencies and let the user select the reference reduction coefficient Ks to be used. The reference reduction coefficient Ks is stored in a predetermined area in the internal storage device 200 (FIG. 1) in a table format.

The reduction coefficient Kf is calculated using the reference reduction coefficient Ks described above. FIG. 7 is an explanatory diagram showing an example of a method of calculating the reduction coefficient Kf using the reference reduction coefficient Ks. When the skin color representative point FP is located on the reference straight line SL, the value of the reference reduction coefficient Ks corresponding to the position of the skin color representative point FP on the reference straight line SL is the reduction coefficient Kf. On the other hand, when the skin color representative point FP is not located on the reference straight line SL, the reduction coefficient Kf is calculated using the two reference reduction coefficients Ks. These two reference reduction coefficients Ks are the reference reduction coefficients Ks along the two reference straight lines SL that are closest to the skin color representative point FP. In the example of FIG. 7, the two reference lines SL closest to the skin color representative point FP are the orange reference line SLo and the red reference line SLr. Therefore, the reference reduction coefficient Ks used for calculating the reduction coefficient Kf is the orange reference reduction coefficient Kso and the red reference reduction coefficient Ksr.

First, a point RP that is a point on two reference straight lines SL and whose distance Lt from the ideal point IP is equal to the distance between the ideal point IP and the skin color representative point FP (hereinafter referred to as "reference point RP"). Calculate the reference reduction coefficient Ks in. Figure 7 shows the reference points RP (RPo and RPr) on the two reference lines SL (SLo and SLr). FIG. 8 is an explanatory diagram showing an example of the reference reduction coefficient Ks at the reference point RP. FIG. 8 (a) shows the reference reduction coefficient Kso (r) at the reference point RPo on the orange reference line SLo, and FIG. 8 (b) shows the reference reduction coefficient at the reference point RPr on the red reference line SLr. The coefficient Ksr (r) is shown.

Next, the reduction coefficient Kf is calculated by weighting and averaging the reference reduction coefficient Ks of the two reference point RPs according to the distance between the reference color fog point SP corresponding to the reference point RP and the skin color representative point FP. .. As shown in Fig. 7, if the distance between the orange reference color cast point SPo and the skin color representative point FP is Lo and the distance between the red reference color cover point SPr and the skin color representative point FP is Lr, the reduction coefficient Kf is , Calculated by the following equation (1).

<maths num="1"><img file="JP4371069B2_D0001.tif" /></maths>

As described above, the reduction coefficient Kf is calculated using the reference reduction coefficient Ks. In this embodiment, since the reference reduction coefficient Ks is set to be a value within the range of 0 to 0.8, the calculated reduction coefficient Kf is also set to a value within the range of 0 to 0.8.

In this embodiment, the value of the reduction coefficient Kf is the value of the skin color representative point FP (a).^*And b^*), It is set variably. Here, the value of the skin color representative point FP represents the tendency of color cast such as the direction and degree of color cast of the target image. Therefore, the reduction coefficient Kf, that is, the value of the ratio of the standard correction amount Δ (N) to the difference between the value of the skin color representative point FP and the value of the ideal point IP is variably set according to the tendency of color cast. .. Further, the color balance correction is performed using the execution correction amount Δ (E) set based on the standard correction amount Δ (N). Therefore, in the image processing apparatus 100 of this embodiment, preferable color balance correction can be performed according to the tendency of color cast.

Further, in this embodiment, since the tendency of each reference reduction coefficient Ks can be made different for each reference straight line SL, the value of the reference reduction coefficient Ks depends on the direction of the color cast even if the color cast is the same. Can be different. Since the reduction coefficient Kf is calculated using the reference reduction coefficient Ks, the value of the reduction coefficient Kf can also be different depending on the direction of color cast. Therefore, in the image processing apparatus 100 of this embodiment, preferable color balance correction can be performed according to the direction of color cast.

Further, in this embodiment, the value of the reference reduction coefficient Ks along one reference straight line SL can be set differently according to the distance from the ideal point IP, so that the reference reduction coefficient Ks depends on the degree of color cast. The values of can be different. Since the reduction coefficient Kf is calculated using the reference reduction coefficient Ks, the value of the reduction coefficient Kf can also be different depending on the degree of color cast. Therefore, in the image processing apparatus 100 of this embodiment, preferable color balance correction can be performed according to the degree of color cast.

Further, in this embodiment, since the reference reduction coefficient Ks is calculated using the preset reference reduction coefficient Ks, the reduction coefficient Kf can be easily calculated and the processing speed can be increased. it can. In addition, the tendency of the reduction coefficient Kf according to the tendency of color cast can be easily set, and the setting for performing preferable color balance correction can be easily performed.

B. Modification example: The present invention is not limited to the above-mentioned examples and embodiments, and can be carried out in various embodiments without departing from the gist thereof. For example, the following modifications can be made.

B1. Modification 1: In the above embodiment, the reduction coefficient Kf is calculated from the converted skin color representative value FVc using the reference reduction coefficient Ks, but the converted skin color representative value FVc (or skin color representative value FV) and the reduction coefficient Kf are calculated. The relationship with and may be calculated in advance and held as a table. In this way, the speed of the color balance correction process can be increased.

B2. Modification 2: The calculation method of the reduction coefficient Kf using the reference reduction coefficient Ks shown in the above embodiment is only an example, and another method can be adopted as the calculation method of the reduction coefficient Kf using the reference reduction coefficient Ks. Is. For example, in the above embodiment, the reduction coefficient Kf is calculated using two of the set six reference reduction coefficients Ks, but only one reference reduction coefficient Ks may be used, or three or more. The reduction coefficient Kf may be calculated by using the reference reduction coefficient Ks. Further, in the above embodiment, when calculating the reduction coefficient Kf, the distance between the skin color representative point FP and the reference color fog point SP is used for weighting averaging, but the distance between the skin color representative point FP and the reference point RP is used. May be used for weighted averaging. Further, in the above embodiment, six reference reduction coefficients Ks are set, but the number of the reference reduction coefficients Ks to be set is arbitrary, and five or less reference reduction coefficients Ks may be set. , 7 or more reference reduction coefficients Ks may be set.

Further, the value of the reference reduction coefficient Ks shown in the above embodiment is merely an example, and the value of the reference reduction coefficient Ks may be set to a different value. For example, in the above embodiment, the reference reduction coefficient Ks is set to a value in the range of 0 to 0.8, but the reference reduction coefficient Ks may be set to a value in a different range such as 0 to 1.0.

B3. Modification 3: In the above embodiment, as a plane representing hue and saturation, a^*-b^*A plane is used, but u is used as a plane to represent hue and saturation.^*-v^*Different planes such as planes may be used.

B4. Modification 4: In the above embodiment, the face region is used as the human skin region represented in the image, but a region other than the face may be used as the human skin region.

<figref num="1">Explanatory drawing which shows schematic structure of the image processing apparatus as an Example of this invention.</figref><figref num="2">The flowchart which shows the flow of the color balance correction processing by the image processing apparatus of an Example.</figref><figref num="3">Explanatory drawing which shows the outline of the calculation method of the standard correction amount Δ (N).</figref><figref num="4">Explanatory drawing which shows an example of the setting method of execution correction amount Δ (E).</figref><figref num="5">Explanatory drawing which shows an example of a reference straight line SL on a * -b * plane.</figref><figref num="6">Explanatory drawing which shows an example of a reference reduction coefficient Ks.</figref><figref num="7">Explanatory drawing which shows an example of the calculation method of the reduction coefficient Kf using the reference reduction coefficient Ks.</figref><figref num="8">Explanatory drawing which shows an example of a reference reduction coefficient Ks in a reference point RP.</figref>

Code description

100 ... Image processing device 110 ... CPU 120 ... Display 130 ... Operation unit 140 ... external storage 150 ... Interface section 160 ... Bus 200 ... internal storage 210 ... Color balance correction unit 212 ... Face area input section 213 ... Face area detector 214 ... Representative value calculation unit 216 ... Correction processing unit 217 ... Reduction coefficient calculation unit 218 ... Correction amount calculation unit 300 ... Digital Still Camera 400 ... Printer