JPH0837667A - Color correction device - Google Patents

Abstract

PURPOSE:To correct colors corresponding to the different kinds of light sources and to position a light source position by a light source estimation method for which chromatic colors and achromatic colors within a screen are paid attention to by dividing the screen into plural, obtaining the light source positions for respective areas and performing interpolation on the screen. CONSTITUTION:First, the screens are divided into a lot of small areas so as to obtain the color average values of (r) (red,) (g) (green) and (b) (blue.) For instance, the small areas are constituted by dividing the screen into 6X8, the color average values of the respective small areas are obtained in an average value computing part 1 and color difference values are obtained by a prescribed formula by a color conversion part 2 by using the color average values. That is, it is defined that X=(b-y)/y; Y=(r-y)/y, X and Y are defined as the color difference values, (y) is defined as luminance and the screens are divided into plural so as to decide the light source position of color space. For instance, division is defined as quadri-section, the color difference values of the respective small areas are outputted to intra-frame light source decision parts 3-6, the light source positions are obtained by them and the light source positions of respective picture elements are obtained based on the light source positions of respective divided frames by an interpolation processing part 7.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a color correction device for color correction such as a video camera or a still camera.

[0002]

2. Description of the Related Art The human eye has the property of color adaptation, and the colors of the same object appear to be almost the same even under various light sources. However, since video cameras do not have color adaptation, color reproduction may be different even for the same object depending on the type of light source. Therefore, auto white balance correction is essential for video cameras and the like, and various methods have been proposed conventionally.

Here, a conventional color image pickup device will be described. FIG. 13 is an overall configuration diagram of a conventional color image pickup apparatus. Reference numeral 101 is an image pickup element, 102 is a color separation unit, 103 is a gain adjustment unit, 104 is a white balance calculation unit, and 105.
Is a color conversion unit, and 106 is an encoder. FIG. 14 shows the details of the white balance calculation unit 104 in FIG. 13, where 107 is an average value calculation unit, 108 is a color conversion unit,
Reference numeral 109 is a light source selection unit, and 110 is a gain calculation unit.

The operation of the color image pickup device constructed as described above will be described below. First, the image sensor 10
The image signal obtained by 1 is r by the color separation unit 102.
It is separated into gb basic color signals. Then, the rgb signal is gain-controlled by the gain adjustment unit 103, becomes an r′g′b ′ signal, and is sent to the color conversion unit 105. Here, from the r′g′b ′ signal output from the gain adjusting unit 103, r−y,
Generate a by color difference signal. The color difference signal is input to the encoder 106, where it is converted into a standard color signal of the NTSC system or the like, and then output.

The color signal rgb output from the color separation unit 102 is also input to the white balance calculation unit 104. In the white balance calculation unit 104, the average value calculation unit 107 first averages the rgb image signals for each screen, and the color conversion unit 108 converts the rgb image signals into color difference signals r-y and by. In the light source selection unit 109, on the color difference plane, as shown in FIG.
The light source selection area 111 determined by various light source positions as shown in FIG. 6 is compared with the obtained position of the input image, and the subject is illuminated depending on whether or not the input image is within the light source selection area 111. Select a light source type expected to be.
Then, the gain calculation unit 110 outputs the gain value corresponding to the estimated light source, and the gain adjustment unit 103 outputs r.
The gain of the gb signal is controlled to white balance the image signal.

[0006]

In the conventional structure as described above, it is premised that the light source selection unit has a single light source, and cannot be corrected when illuminated by a plurality of different light sources. was there.

When there are a plurality of different light sources, two cases are conceivable: a case where lights are mixed to illuminate the screen and a case where different lights illuminate different areas in the screen. Especially in the text, the latter is called a heterogeneous light source. In color correction, color correction in the case of different light sources is the most difficult problem to solve. For example, when the halogen light (color temperature near 3000K) and the sunlight (near daylight 5000K) exist in different light source states as shown in FIG. 16, the area illuminated by the halogen light is red in the conventional method, and is illuminated by the sunlight. The created area is reproduced in blue, often resulting in an unnatural image.

In view of the above point, the present invention has an object to provide a color correction device capable of performing correction even when different types of light sources are used.

[0009]

In order to achieve the above object, the present invention divides the entire screen into a large number of small areas and calculates a color average value from the density values of the pixels in each small area. An in-frame light source determination unit that determines a light source position on the color space for each divided frame from the color average value by providing a smaller number of divided frames on the screen than the small area, and interpolates the light source position of the divided frame. According to the configuration, an interpolation processing unit that determines the light source position of each pixel on the screen and a gain calculation unit that outputs a gain for correcting the density value of each pixel from the obtained light source position of each pixel are configured. is there.

Further, a multi-linear function is used in the interpolation process for determining the light source position of each pixel on the screen.

Further, the entire screen is divided into a large number of small areas, and an average value calculating section for calculating a color average value from the density values of pixels in each small area, and the color average value for hue, saturation, and luminance. The color conversion unit for converting into three-dimensional information and the light source positions of a plurality of illuminations which are known in advance as the principal axis direction of the color distribution of the small area on the plane constituted by the hue and the saturation are actually illuminated in the image. A main axis selection unit that determines the light source position that has been set, and a gain calculation unit that outputs a gain for correcting the density value of each pixel from the obtained light source position.

Further, on a plane constituted by hue and saturation, a main axis for determining the light source position actually illuminated in the image based on the color distributed in the vicinity of the axis passing through the light source position known in advance. This is a configuration including a selection unit.

Further, the entire screen is divided into a plurality of small areas, and an average value calculating section for calculating a color average value from the density values of pixels in each small area, and the color average value for hue, saturation, and luminance. On the plane formed by the color conversion unit for converting into three-dimensional information and the hue and saturation, the light source positions of a plurality of illuminations which are known in advance as the main axis direction of the color distribution are actually illuminated in the image. A main axis selection unit that determines a light source position, a first reliability determination unit that outputs a reliability that indicates its certainty, and a light source position that is determined by the distance between the color average value and a light source position that is known in advance. A light source selection unit, a second reliability determination unit that outputs a reliability that represents the certainty thereof, and a light source determination unit that determines the light source position based on the reliability of each of the first and second reliability determination units, Correct the density value of each pixel from the obtained light source position of each pixel A structure provided with the gain calculation unit for outputting a gain fit.

Further, the in-frame light source determination section includes a color conversion section for converting the color average value into three-dimensional information of hue, saturation, and luminance, and a color distribution within the divided frame on a plane composed of hue and saturation. The main axis selecting unit that determines the position of the light source that is actually illuminated in the image from the light source positions of a plurality of illuminations that are known in advance as the main axis direction of the, and the first reliability that outputs the reliability that indicates the certainty. A determination unit, a light source selection unit that determines the light source position on the condition of the distance between the color average value and a light source position that is known in advance, a second reliability determination unit that outputs a reliability that represents the certainty thereof, This is a configuration using a light source determination unit that determines the light source position based on the reliability of each of the first and second reliability determination units.

[0015]

According to the present invention, a plurality of light source positions are obtained by providing the above means to determine the light source position for each divided frame in the screen. By performing interpolation of the obtained light source position, different color correction is realized for each pixel in the screen.

In order to obtain the light source position for each divided frame more accurately, the light source position is determined based on the principal axis direction of the chromatic color distribution on the color space.

Furthermore, the accuracy of deriving the light source position is improved by integrating the light source estimation method with emphasis on achromatic colors and the light source estimation method with emphasis on chromatic colors.

[0018]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a first embodiment of a color correction apparatus according to the present invention. An embodiment of the entire configuration using the present invention is shown in FIG.
Is the same as In this embodiment, the white balance calculation unit 10
4 regarding the internal configuration.

In FIG. 1, reference numeral 1 denotes an average value calculation unit that divides the screen into a large number of small areas, and outputs color average values of r (red), g (green), and b (blue). Value ((by) / y, (ry)
/ y) is a color conversion unit that converts the color difference value into three, four, five, and six division frames that divide the screen into four frames that determine the light source position from the color difference value in each division frame. Internal light source determination unit,
Reference numeral 7 is an interpolation processing unit that determines the light source position of each pixel on the screen by interpolating the light source position of each divided frame obtained from the in-frame light source determination units 3, 4, 5, and 6, and 8 is an interpolation processing unit 7. It is a gain calculation unit that calculates a gain (bcont, rcont) from the obtained color difference value of each pixel.

In the present embodiment, the effect of recognizing light having different spectral distributions as colors that are quite close to each other is realized engineeringally. This is a method of determining the light source position for each area in the image and realizing color correction in accordance with the light source. The light source position (white position) on the color difference plane is changed according to the position in the screen.

First, in order to obtain the color average values of r, g, and b,
Divide the screen into many small areas. For example, as shown in FIG. 2A, the screen is divided into 6 × 8 to form a small area. The average value calculator 1 obtains the color average value of each small area. Color converter 2
Then, the color difference value is obtained as in (Equation 1) using the color average value.

[0022]

[Equation 1]

The screen is divided into a plurality of parts in order to determine the light source position in the color space. For example, it is divided into four as shown in FIG. The color difference value of each small area is output to the in-frame light source determination units 3 to 6. The in-frame light source determining units 3, 4, 5, and 6 use, for example, conventional light source positions ((X1, Y1), (X2, Y2), (X3, Y3), (X
4, Y4)) is required. The interpolation processing unit 7 obtains the light source position of each pixel on the screen by interpolation based on the light source position of each divided frame. As the interpolation processing, a multiple linear function as shown in FIG. 2C in which light source positions are arranged at four corners of the frame as shown in FIG. 2B is configured. When expressed by an equation, it becomes as shown in (Equation 2).

[0024]

[Equation 2]

Color difference ((b-
(y) / y, (ry) / y) and ((bo-yo) / yo,
(ro-yo) / yo) is calculated, and the gain (bcon
t, rcont) is calculated. This operation is performed for all pixels in one frame.

FIG. 3 (a) shows an image when sunlight is illuminated on the left side of the screen and halogen light is illuminated on the right side, and FIG. 3 (b) shows the light source position on the color difference plane at that time. There is. In the in-frame light source determining unit, the two in-frame light source determining units arranged on the left side of the frame estimate sunlight as the light source, and the two in-frame light source determining units disposed on the right side estimate the halogen light as the light source. Then, in the entire screen, the light source position obtained by the multiple linear function is assigned to each pixel.

As described above, in the present embodiment, the light source position is determined for each divided frame in the screen and is interpolated on the screen, which is a correction method applicable to different light sources.

FIG. 4 shows a second color correction apparatus according to the present invention.
3 is a block diagram showing the configuration of the embodiment of FIG. An example of the entire configuration using the present invention is the same as that of the conventional example shown in FIG. The present embodiment relates to the internal configuration of the white balance calculation unit 104. In FIG. 4, 9 is an average value calculator that divides the screen into a large number of small areas and outputs color average values of r (red), g (green), and b (blue), and 10 is an average value (( by) / y, (r-
A color conversion unit that converts y) / y) into color difference values, 11 is a main axis selection unit that determines the light source position based on the main axis of the color distribution of the color difference values, 12
Is a gain (b from the color difference value of each pixel obtained by the main axis selection unit 11).
cont, rcont) is a gain calculator.

The conventional example is a system in which the light source position where the distribution of color difference values is most concentrated on the color difference plane is used as the illumination light source. However, there are many images in which the color difference values are not concentrated on any of a plurality of light source positions that are known in advance. Therefore, the present embodiment proposes a light source estimation method using chromatic colors.

First, the principle for estimating the light source will be described with reference to FIG. In many cases, the light reflected from the object surface 13
It has been conventionally known that decomposition into two additive components, specular reflection Ys (θ, λ) and diffuse reflection Yd (θ, λ). When expressed by an equation, it becomes as shown in (Equation 3).

[0031]

(Equation 3)

Here, the geometrical parameter and the wavelength component can be separated as in (Equation 4).

[0033]

[Equation 4]

Further, the power spectrum distribution of the reflected light is decomposed into the product of the spectral reflectance and the power spectrum distribution of the incident light as shown in (Equation 5).

[0035]

(Equation 5)

In practice, it has been reported that the surface of an inhomogeneous object is oily and the refractive index of light is constant with wavelength. If the refractive index is constant in the visible range, the specular component does not depend on the wavelength and can be expressed as in (Equation 6).

[0037]

(Equation 6)

Therefore, the reflected light Y (θ, λ) is the power spectrum distribution E (λ) of the incident light and its diffusion component Sd (λ) E (λ).
Is the linear sum of. If this is expressed on the color difference plane, the color distribution is concentrated on a straight line between the color of the light source that is the incident light and the color of the object that is the diffuse reflection component. FIG. 6 shows a case where a red subject is illuminated with a fluorescent lamp. 14 is the distribution of color difference values is 1
The dotted line 5 indicates the main axis of the color difference value distribution, and 16 indicates the light source position on the color difference plane of the fluorescent lamp. Although the saturation of the red subject is constant, there is a change in the geometrical parameter (change in luminance), so that the color is linearly distributed between the light source and the object color on the color difference plane.

From the above knowledge, in this embodiment, when the colors are concentrated in one color, the principal axis direction of the color distribution is obtained, and the condition of intersecting with any one of the light source positions known in advance is used to obtain the color difference plane of the illumination light source. Estimate the light source position above.
Generally, for the main axis direction of the distribution, the position and inclination of the main axis can be obtained by using, for example, the least squares method of the multivariate analysis method. However, in view of the necessity of performing the real-time processing, a more simplified method is adopted in this embodiment.

That is, instead of directly obtaining the principal axis of the color distribution, an attempt is made to construct a simple color model that limits the principal axis direction from the light source position. The models are limited to red such as sunset, green of trees, skin color (between red and yellow), and blue of the sky. These models are constructed by parameters of subject color and light source position. The model is represented by a region on the color difference plane as shown in FIG. Reference numeral 17 is a light source position, 18 is a fan-shaped area representing red, 19 is flesh color, 20 is green, and 21 is blue. The larger the number of color difference values belonging to the fan-shaped region, the higher the possibility of being the model. A red model corresponding to each light source position is shown in FIG.

In FIG. 8, 22 is a color temperature of 2500K, 23
Is 3100K, 24 is 4300K, 25 is 5400K, 26 is 7000
K, 27 are white fluorescent lamps, 28 is a three-wavelength daylight fluorescent lamp,
29 is a light source position of a daylight color fluorescent lamp. The other three colors are similarly configured by changing the angle of the area on the fan with respect to the light source position. When expressed by an equation, it becomes as shown in (Equation 7).

[0042]

(Equation 7)

The light source is estimated using this model. That is, when the light source is li, the probability that the subject color cj exists in the light source li is represented by the number of pixels N. By calculating the number of pixels N for each light source and each subject color, it is possible to derive which subject color of which light source is likely to exist.

An example of the algorithm will be shown below. [STEP1] The frame is divided into a plurality of parts, and the average value of color signals (color difference signals in this case) (called color difference average value) is obtained. (6
* In the case of 8 divisions, the average value of 48) [STEP2] From the average value of color difference, as a light source based on (Equation 7) (2500K, 3100K, 4300K, 5400K,
7,000K, white, 3 wavelength type daylight color, daylight color)
As a color model, a total of 32 types of pixel numbers N are obtained for four colors (red, flesh color, green, blue). [STEP3] Select the light source that derives the maximum number of pixels for each color. Since Nmax (li, cj) in (Equation 8) has four colors, there are four.

[0045]

(Equation 8)

[STEP4] The position of the light source is estimated using the number of pixels for the selected light source as a weighting coefficient.

[0047]

[Equation 9]

FIG. 9A shows an example of an image in which a large face is imaged on the screen and a lot of skin-colored areas 30 are present.
When a white fluorescent lamp is used for illumination as a light source, a color difference value 33 exists in a fan-shaped area 32 representing the skin color at the light source position 31 of the white fluorescent lamp as shown in FIG. 9B. Color difference values also exist in fan-shaped areas provided at other light source positions,
The flesh color area 32 of FIG. 9B has the largest number of color difference values, and as a result, a white fluorescent lamp is selected as the light source.

As described above, in the present embodiment, the light source is estimated from the change in the chromatic color of the object due to the illumination, so that the light source can be accurately estimated even if the number of division frames increases and color deviation occurs.

Next, a third embodiment of the present invention will be described with reference to the drawings. FIG. 10 is a block diagram showing the configuration of the third embodiment of the color correction apparatus according to the present invention. An example of the entire configuration using the present invention is similar to that of the conventional example shown in FIG. The present embodiment relates to the internal configuration of the white balance calculation unit 104. In FIG. 10, reference numeral 34 divides the screen into a large number of small areas, r (red), g (green), and b.
An average value calculator that outputs the color average value of (blue), 35 is a color conversion unit that converts the color average values ((by) / y, (ry) / y) into color difference values, and 36 is the color distribution of the color difference values. The main axis selecting section for determining the light source position based on the main axis of 37, 37 is the first reliability determining section for outputting the reliability indicating the certainty of the light source position obtained by the main axis selecting section 36, and 38 is the conventional example. A light source selection unit having a similar function, 39 is a second reliability determination unit that outputs the reliability that indicates the certainty of the light source position obtained by the light source selection unit 38, 40
Is a light source determination unit that combines the light source positions obtained by the spindle selection unit 36 and the light source selection unit 38 based on the respective reliability levels output by the first and second reliability determination units, and 41 is a light source determination unit 40. It is a gain calculation unit that calculates the gain (bcont, rcont) from the obtained color difference value of the light source position.

The present embodiment is a system in which both the chromatic color and the achromatic color on the screen are used to determine the light source position with higher accuracy. First, in order to obtain the color average value of r, g, b,
The screen is divided into many small areas. For example, as shown in FIG. 2A, the screen is divided into 6 × 8 to form a small area. The average value calculator 34 obtains the color average value of each small area. In the color conversion unit 35, the color difference value is obtained by using the color average value as in (Equation 10).

[0052]

[Equation 10]

In the spindle selecting section 36, the light source position is obtained by the following steps using the model as shown in (Equation 11) as in the second embodiment.

[0054]

[Equation 11]

The light source is estimated using this model. That is, when the light source is li, the probability that the subject color cj exists in the light source li is represented by the number of pixels N. By calculating the number of pixels N for each light source and each subject color, it is possible to derive which subject color of which light source is likely to exist.

An example of the algorithm will be shown below. [STEP1] The frame is divided into a plurality of parts, and the average value of color signals (color difference signals in this case) (called color difference average value) is obtained. (6
* In the case of 8 divisions, the average value of 48) [STEP2] From the average value of color difference, as a light source based on (Equation 11) (2500K, 3100K, 4300K, 5400K,
7,000K, white, 3 wavelength type daylight color, daylight color)
As a color model, a total of 32 types of pixel numbers N are obtained for four colors (red, flesh color, green, blue). [STEP3] Select the light source that derives the maximum number of pixels for each color.

[0057]

(Equation 12)

[STEP 4] The position of the light source is estimated using the number of pixels for the selected light source as a weighting coefficient.

[0059]

(Equation 13)

The first reliability determining section 37 determines the reliability of the light source position under the following conditions. In (Equation 12), there are four types of light sources that derive the maximum number of pixels. For example, as shown in FIG. The condition here is the maximum number of pixels, the ratio of the maximum number of pixels in a light source different from the light source that has reached the maximum number of pixels, and the reliability of the accuracy of the light source position based on the ratio of the maximum number of pixels to the total number of pixels. Make up degrees.

[0061]

[Equation 14]

On the other hand, the light source selection unit 38 determines the light source by the method shown in the conventional example. (Equation 15), (Equation 1)
6).

[0063]

(Equation 15)

[0064]

[Equation 16]

The light source li having the maximum number of pixels in (Equation 16) serves as an illumination light source. Furthermore, the reliability representing the certainty of the light source position obtained by the equation (16) is shown by the equation (17).

[0066]

[Equation 17]

The light source position is determined from the reliability R1 and R2. The larger R1 and R2, the higher the reliability. Therefore, the following processing can be performed to determine the light source. (1) When R1> k1 and R2 <k2, spindle selection unit 36
(2) When R1 <k1 and R2> k2, the light source selection unit 38 is used.
(3) R1> k1 and R2> k2 or R1 <k1 and R
When 2 <k2, the intermediate value between the light source positions of the spindle selection unit 36 and the light source selection unit 38 is set as the light source position.

An example of the effect is briefly shown in FIG. 44
Reference numerals 46 and 46 are light source selection areas, 43 and 45 are fan-shaped areas for obtaining the principal axes of colors, and 42 and 47 are distributions of color difference values.

FIG. 12A shows the case where the colors are biased.
(b) is a case where various colors including achromatic colors exist.
According to the present embodiment, in the case of FIG. 12A, the reliability R1 is large and the reliability R2 is small, and the light source position of the spindle selection unit 36 is adopted. On the other hand, in the case of FIG. 12B, the reliability R1
Is small and the reliability R2 is large, the light source selection unit 38
The light source position of is adopted. Thus, the light source can be accurately obtained according to the state of the image.

Further, in the first embodiment, in order to perform color correction on a pixel-by-pixel basis, it is necessary to divide the screen into a plurality of parts and obtain the origin position of color correction for each. However, the division frame naturally has a smaller area than the entire screen, and as a result, the color deviation in the division frame becomes large, and there are cases in which sufficient accuracy cannot be expected with the conventional method. However, in the third embodiment, the spindle selection unit 36, the first reliability determination unit 37, the light source selection unit 38, the second reliability determination unit 39, and the light source determination unit 40 are included in each frame light source determination unit 3, If used for 4, 5 and 6,
It is possible to realize a color correction device that can sufficiently cope with color deviation in the divided frame.

[0071]

As described above, according to the present invention, it is possible to perform color correction corresponding to different light sources by dividing the screen into a plurality of parts, obtaining the light source position for each, and interpolating on the screen. In addition, the light source position can be accurately obtained by using the light source estimation method focusing on the chromatic color and the achromatic color on the screen.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a color correction apparatus according to a first embodiment of the present invention.

2A to 2C are diagrams showing screen division and light source position interpolation processing.

3A and 3B are diagrams showing states of different light sources.

FIG. 4 is a block diagram showing a configuration of a color correction apparatus according to a second embodiment of the present invention.

FIG. 5 is a diagram showing light reflection.

FIG. 6 is a diagram showing light reflection characteristics of a chromatic substance.

FIG. 7 is a diagram showing light source estimation using chromatic colors.

FIG. 8 is a diagram showing light source estimation using chromatic colors.

9A and 9B are diagrams showing light source estimation using chromatic colors.

FIG. 10 is a block diagram showing the configuration of a color correction apparatus according to a third embodiment of the present invention.

FIG. 11 is a diagram showing a third embodiment.

12 (a) and 12 (b) are views showing the effect of the third embodiment.

FIG. 13 is a block diagram showing the overall configuration of a conventional color correction device.

FIG. 14 is a block diagram showing another conventional color correction device.

FIG. 15 is a diagram showing an operation of a conventional color correction device.

FIG. 16 is a diagram for explaining a problem to be solved by the invention.

[Explanation of reference numerals] 1, 9, 34 Average value calculation unit 2, 10, 35 Color conversion unit 3, 4, 5, 6 In-frame light source determination unit 7 Interpolation processing unit 8, 12, 41 Gain calculation unit 11, 36 Spindle Selector 13 Material surface 14, 33, 42, 47 Color difference value distribution 15 Main axis 16, 17 Light source position 18, 19, 20, 21, 32, 43, 45 Fan-shaped area 22, 23, 24, 25, 26, 27 , 28, 29, 3
DESCRIPTION OF SYMBOLS 1 Light source position 30 Skin color area 37 1st reliability determination part 38 Light source selection part 39 2nd reliability determination part 40 Light source determination part 44, 46 Light source determination region

Claims (6)

[Claims] 1. An average value calculation unit that divides the entire screen into a plurality of small areas and calculates a color average value from the density values of pixels in each small area, and a number of division frames smaller than the small areas on the screen. And an inter-frame light source determining unit that determines the light source position on the color space for each divided frame from the color average value, and an interpolation that determines the light source position of each pixel on the screen by interpolating the light source position of the divided frame. A color correction apparatus comprising: a processing unit; and a gain calculation unit that outputs a gain for correcting the density value of each pixel from the obtained light source position of each pixel. 2. A multiple linear function is used for interpolation processing for determining the light source position of each pixel on the screen.
The described color correction device. 3. An average value calculation unit that divides the entire screen into a plurality of small areas and calculates a color average value from the density values of pixels in each small area, and the color average value of hue, saturation, and luminance. The color conversion unit for converting into three-dimensional information and the light source positions of a plurality of illuminations which are known in advance as the principal axis direction of the color distribution of the small area on the plane constituted by the hue and the saturation are actually illuminated in the image. A color correction apparatus comprising: a main axis selection unit that determines a light source position that is set, and a gain calculation unit that outputs a gain for correcting the density value of each pixel from the obtained light source position. 4. A principal axis for determining a light source position actually illuminated in an image based on a color distributed in the vicinity of an axis passing through a light source position which is known in advance on a plane constituted by hue and saturation. The color correction apparatus according to claim 3, further comprising a selection unit. 5. An average value calculation unit that divides the entire screen into a plurality of small areas and calculates a color average value from the density values of pixels in each small area, and the color average values of hue, saturation, and luminance. On the plane formed by the color conversion unit for converting into three-dimensional information and the hue and saturation, the light source positions of a plurality of illuminations which are known in advance as the main axis direction of the color distribution are actually illuminated in the image. A main axis selection unit that determines a light source position, a first reliability determination unit that outputs a reliability that indicates its certainty, and a light source position that is determined by the distance between the color average value and a light source position that is known in advance. A light source selection unit, a second reliability determination unit that outputs a reliability that represents the certainty thereof, and a light source determination unit that determines the light source position based on the reliability of each of the first and second reliability determination units, To correct the density value of each pixel from the obtained light source position of each pixel Color correction apparatus characterized by comprising a gain calculation unit for outputting obtained. 6. The in-frame light source determining unit converts a color average value into three-dimensional information of hue, saturation, and luminance, and a color distribution in a divided frame on a plane composed of hue and saturation. The main axis selecting unit that determines the position of the light source that is actually illuminated in the image from the light source positions of a plurality of illuminations that are known in advance as the main axis direction of the, and the first reliability that outputs the reliability that indicates the certainty. A determination unit, a light source selection unit that determines the light source position based on the distance between the color average value and the light source position that is known in advance,
A second reliability determination unit that outputs a reliability that represents the certainty and a light source determination unit that determines a light source position based on the reliability of each of the first and second reliability determination units. The color correction device according to claim 1.