JP3052592B2 - Light source estimation device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a control and white balance adjustment apparatus for reproducing correct tone in a video camera line of it, to an optical source estimating apparatus.

[0002]

2. Description of the Related Art In recent years, with the spread of video cameras, means for following shooting conditions in order to obtain beautiful colors has become increasingly important. In particular, it is very important to accurately estimate the light source at the time of shooting in order to make color reproduction accurate.

Conventionally, as means for estimating a light source at the time of photographing, there have been known means for estimating using a plurality of light-sensitive elements having different spectral characteristics and means for estimating through image processing. . In general, the light source estimating apparatus has shifted from a means using a photosensitive element to a means for processing and estimating an image signal (hereinafter, referred to as AWB) due to restrictions such as design cost.

[0004] The basis of the method of estimating the light source by image processing is based on the assumption that mixing and averaging the colors of the objects on the screen results in an achromatic color (Gray World Assumption).
ion: GWA). If the screen contains a sufficiently large number of colors, this assumption holds. If the color information obtained from the image signal is mixed and averaged, it should represent the color of the light source at the time of shooting. As a method of performing color adjustment based on this assumption, a method described in JP-A-56-36291 is known.

[0005] However, the conventional method based on the GWA has a drawback that the light source estimation is erroneous when a specific color occupies a wide area of the screen.

The following two methods can be considered to solve this drawback. First, GWA is established by a configuration including a variance calculation unit that calculates the variance of the color distribution of the screen, and a threshold processing unit that performs a threshold process based on whether an output from the variance calculation unit is larger than a set value. The failure is determined for each image, and the adoption / non-adoption of the light source estimation result is controlled.

Second, considering the power-on time when it is necessary to perform light source estimation for all images, G
Instead of the light source estimation based on the WA, a method of estimating the light source by performing an operation of extracting a dominant color on the screen from all images can be considered.

By using these techniques, it is possible to reduce the influence of high-saturation colors and to improve the accuracy of light source estimation.

[0009]

However, the hay is
Like shooting a subject covering a wide area outdoors
The dominant color (color of hay) on the screen is the color of the outdoor light source
However, if it is not an indoor light source, turn on the power for a while.
For example, there is a problem that the light source estimation is wrong. Also the screen
For example, when there are multiple colors that are not very vivid on the top
Looking at the color distribution of the whole image like
Where there is no dominant color even though it looks like an image
In many cases, there is a problem that the light source estimation is wrong.
Was .

[0010]

[0011]

The present invention solves the above-mentioned conventional problems, and is applied to a case where a dominant color does not exist on a screen or an outdoor light source.
When the area occupied by colors that are neither indoor nor light sources is large
It is another object of the present invention to provide a light source estimating apparatus that estimates a light source chromaticity with high accuracy .

[0013]

[0014]

[0015]

In order to achieve this object, a light source estimating apparatus converts an image signal into an appropriate sampling rule.
Sampling circuit that converts according to the rule, and sampling
Converts image signals sampled by the circuit into chromaticity signals
And a chromaticity conversion unit that processes the chromaticity signal to determine a light source chromaticity candidate.
Light source chromaticity candidate calculation unit to be estimated
Dominant color exists by checking the number of chromaticity signals in the limited area
Dominant color existence determination unit that determines whether to perform
Light source chromaticity candidate from candidate calculation unit and light from dominant color judgment unit
Light source color for determining estimated light source chromaticity using source estimation control signal
Divide the chromaticity plane and assign light sources to each area
The light source is determined by the area to which the estimated light source chromaticity belongs.
And a light source estimating unit for estimating.

[0016]

[0017]

[0018]

[0019]

With this configuration, there are many chromaticity signals in the limited area.
If there is a dominant color on the screen
To determine if there is a dominant color on the screen.
Can be. Then, by processing the above determination result,
An image suitable for source estimation can be determined. And
Even when there is no dominant color on the screen
A light source estimator that can suppress errors in
Will be revealed. In addition, whether the dominant color is the color of the light source
If this is not the case, use another method to estimate
By estimating the light source using the light source chromaticity candidate
When the area occupied by a color that is neither a light source nor an indoor light source is large
Light that can effectively perform light source estimation
A source estimation device is realized.

[0020]

[0021]

[0022]

[0023]

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

First, the principle of the first present invention will be described. Conventional AWB is based on the GWA described above. However, when a specific color, especially a highly saturated color, occupies a wide area of the screen, such as in a scene such as a zoom-up of a person wearing red clothes, the average of the colors in the screen is the specific color. It is close to color and not achromatic. As described above, the GWA does not always hold for all scenes. Therefore, in order to realize AWB with better color reproduction, it is necessary to determine whether or not GWA is established for each image, and exclude an image for which GWA is not established from the targets of light source estimation.

Many of the widely used illumination light sources have a color close to blackbody radiation. Based on this, the locus of the color of the blackbody radiation plotted on the chromaticity plane with the temperature of the blackbody as a parameter will be referred to as a “blackbody radiation curve”. Further, considering the simplicity of mounting on an actual device, a straight line that first-order approximates the “blackbody radiation curve” is set on the chromaticity plane, and this will be referred to as a “light source line”. These are shown in FIG.

The statistical variance of the chromaticity data of the image projected in the direction perpendicular to the "light source line" is called "vertical variance". The inventor has set "vertical dispersion" as a parameter for determining whether GWA has been established. On the other hand, the statistical variance of the chromaticity data of the image projected in the direction parallel to the “light source line” will be referred to as “parallel variance”.

The inventor has collected a large number of image data,
The analysis was performed by classifying these into an image in which GWA was not established and the light source estimation was likely to be erroneous, and an image in which GWA was established and light source estimation could be performed correctly. As a result, it became clear that the former (the image in which GWA is not established) has a tendency that the “vertical dispersion” is larger than the latter. On the other hand, as a result of performing the same analysis for “parallel dispersion”, such a tendency was not observed. Therefore, "vertical dispersion" is GW
This is effective as a parameter for determining A.

Therefore, when "vertical dispersion" is large, G
By determining that WA is not established and stopping the light source estimation operation and holding the light source estimation result at that time, it is possible to realize a light source estimation device with reduced errors.

However, such a light source estimation device includes:
There is a problem that the light source estimating operation is stopped at the time of photographing the color chart. Since the color chart is frequently used in a product test of a video camera or the like, it is not preferable to stop the light source estimation for the color chart. Therefore, a device that recognizes whether or not the subject is a color chart is required.

The characteristics of the image obtained by photographing the color chart are as follows.
The following two points can be summarized. One. Many highly saturated colors are distributed. 2. A color with almost the same saturation over all hues,
They should be distributed in approximately equal amounts.

Of the above two features, 1 is "vertical dispersion".
Is an effective index. 2 is an effective index that the color distribution on the chromaticity plane is close to a circle. Therefore, if an image having “large vertical variance and a color distribution close to a circle” is detected, it can be determined that this is a color chart image, and an apparatus that recognizes the color chart can be realized.

The inventor has set a parameter R defined as follows as a parameter for detecting that the distribution of two-dimensional data is close to a circle.

When the two-dimensional data is (x _i , y _i ), R = C (X, Y) where C (X, Y): the correlation coefficient of the two-dimensional data (X _i , Y _i ) _{X i, Y i) :( x} i, 2 -dimensional data is generated from the y _i) as follows: _{X i = a 11 x i +} a 12 y i Y i = a 21 x i + a 22 y i a ₁₁ , A ₁₂ , A ₂₁ , A ₂₂ : A 2 × 2 matrix A component that rotates the coordinate axes so that the “distribution axis” is in the 45 ° direction A ₁₁ = A ₂₂ = (1 + a) / [2 (1 + a ²⁾ )] ^1/2 -A ₁₂ = A ₂₁ = (1−a) / [2 (1 + a ² )] ^1/2 a: slope of “distribution axis” described below “distribution axis”: two-dimensional data The gradient a of the straight line “axis of distribution” where the sum of squares of the distance from each point is the minimum is given by the following equation.

Cov (x, y) a ² + (V (x) −V (y)) a−Cov (x, y) = 0 2Cov (x, y) + V (x) −V (y)> 0 However, Cov (x, y): 2 -dimensional data (x _i, y _i) of covariance V (x): one-dimensional data x _i of variance V (y): dispersion more one-dimensional data y _i is a parameter This is the definition of R. The parameter R defined here is hereinafter referred to as “rotational correlation coefficient”. The “axis of distribution” defined here is the same as that generally known as “principal component axis”.

In calculating the "rotational correlation coefficient", first, the "distribution axis" of the two-dimensional data is extracted. The “distribution axis” is a straight line defined as “a straight line that minimizes the sum of the squares of the distances from each point of the two-dimensional data” as described above. ”Indicates whether or not they are distributed. Next, the coordinate axes are rotated so that this straight line has an inclination 1 (45 ° direction). The correlation coefficient of the two-dimensional data is calculated again in the new coordinate system thus created.
This is the “rotation correlation coefficient”.

The smaller the absolute value of the “rotational correlation coefficient”,
It can be determined that the distribution of the two-dimensional data is close to a circle. Less than,
The reason why the “rotational correlation coefficient” indicates the “flatness” will be described.

The correlation coefficient indicates the strength of the correlation between the chromaticities BY and RY, but does not always indicate the "flatness". For example, when the chromaticity is distributed along the BY axis, the “flatness” is large, but the correlation between BY and RY is low, so the correlation coefficient is small. Therefore, the correlation coefficient is not sufficient as a parameter representing “flatness”.

This is because the chromaticity coordinates are not two-dimensional coordinate values but data representing a two-dimensional amount of color. BY and RY do not originally have a relationship where a correlation exists. Therefore, a statistic for extracting the shape of the distribution of the two-dimensional data is required.

Now, if the "axis of distribution" is well defined, it is possible to extract only the shape of the distribution of the two-dimensional data. In the case of the correlation coefficient, even if the shape of the distribution is the same, the evaluation of the strength of the correlation changes depending on the direction of the “axis of the distribution”, and it is difficult to compare only the shape. "Axis of distribution"
, The shape of the distribution can be evaluated even with the correlation coefficient.

Therefore, 1. 1. Extract "axis of distribution" and
After aligning this in the 45 ° direction (the direction in which the correlation between BY and RY becomes the strongest), 3. By calculating the correlation coefficient again, the "flatness" of the distribution can be evaluated. The above is the reason why the “rotational correlation coefficient” represents the “flatness”.

The present inventor has examined the “rotational correlation coefficient” for a large number of images having a large “vertical variance”. Was found to have an extremely small absolute value. Therefore, if the “vertical dispersion” and “rotational correlation coefficient” of the chromaticity data are comprehensively determined, the color chart can be recognized with extremely high accuracy. In addition, since there is a remarkable difference in the distribution range of the “rotational correlation coefficient” between the image of the everyday feature and the image of the color chart, as shown in FIG. Can be set very easily.

Although the same processing can be performed without using the absolute value of the "rotational correlation coefficient", the processing is simpler when the absolute value is used. Below, for simplicity of explanation,
The absolute value of the “rotational correlation coefficient” will be described simply as “rotational correlation coefficient”. Also, instead of "rotational correlation coefficient",
Although it is possible to recognize a color chart using a simple correlation coefficient of the chromaticity coordinates BY and RY, according to the analysis of the present inventor, the recognition accuracy is lower than the method using the “rotational correlation coefficient”. I know it will be.

Also, when calculating the slope of the "axis of distribution", instead of setting up a strict equation as described above and finding a solution,
For example, an approximate value such as a = V (y) / V (x) × (appropriate sign) can be obtained and substituted.
The above is the principle of the first present invention.

Next, the principle of the second invention will be described. The concept of the operation of the light source chromaticity estimating unit according to the present invention is GWA.
Another object of the present invention is to improve light source chromaticity estimation accuracy by excluding high-saturation colors that adversely affect light source estimation based on the chromaticity. In order to estimate the light source while excluding high-saturation colors, the light source may be estimated only from the colors around the achromatic color of the subject on the chromaticity plane. The problem is that the position of the achromatic color of the subject cannot be determined.

According to the data analysis by the present inventors, even in an image in which the GWA is destroyed, the main color is achromatic in most cases. Therefore, the overall average takes a position on the chromaticity plane that is closer to an achromatic color than a highly saturated color. Therefore, when the data is averaged only in a limited area around the overall average, the average (double average) is less affected by the high-saturation color and approaches an achromatic color. Furthermore, when the average (triple average) is obtained in a limited area around the double average, the effect is further increased. By repeating such a procedure, the multiple average becomes an average of almost only achromatic colors, and the accuracy of estimating the light source chromaticity is improved.

According to the analysis result of the multiple average value, the multiple average value is close to the chromaticity of the dominant color on the screen. Therefore, on a screen with many achromatic colors, the multiple average is almost the average of only achromatic colors, and on a screen with many hay colors,
The multiple average is almost the average of only the color of the hay. On the other hand, if there is no dominant color on the screen, the multi-average value will be close to what is most common among the colors around the overall average chromaticity of the image, and what color will be close I can't say for sure. An example of an image in which a dominant color does not exist is an image in which the background is hay grass and the color of the clothes occupies half of the screen by zooming in on a person wearing red clothes. At this time, the color distribution of the image is a distribution of the color of hay and the red color of clothes, and there is no dominant single color. Therefore, it is not known where the multiple average value having a characteristic close to the dominant chromaticity will be.

From these facts, it can be said that the multiple average value is effective when a dominant color exists on the screen. Therefore, when there is a dominant color on the screen, the estimation of the light source chromaticity is performed using the multiple average value. By performing the estimation, the accuracy of estimating the light source chromaticity is improved.

An example of an image for which light source chromaticity estimation based on multiple average values is effective is an image in which a green background of a tree occupies a small area, such as a building, with a color background close to an achromatic color. In this case, the overall average chromaticity of the image data is on the lower color temperature side as compared with the achromatic color, and the multiple average value is close to the achromatic color. Since the achromatic color reflects the color of the light source, correct light source estimation can be performed by using the multiple average value, but incorrect light source estimation is performed by using the total average value.

According to the principle of the first aspect of the present invention, when the vertical variance is large, it is determined that the GWA is not established, the light source estimation operation is stopped, and the light source estimation result at that time is held, whereby an error is detected. It is possible to realize a light source estimating apparatus with reduced noise. " At that time, the relationship between the image and the vertical dispersion was not described. Thus, taking the example of a subject such as a bust shot of a person wearing highly saturated clothes, the effectiveness of this device will be described while analyzing the relationship between the image and the vertical dispersion. In such a subject, the saturation of some of the colors sampled from the screen becomes relatively high, the spread of the sampling data on the chromaticity plane increases, and the vertical dispersion increases. Therefore, inside the device, it is determined that the subject is outside the GWA application limit. On the other hand, when the light source estimation based on GWA is performed on the subject, the average color of the entire screen becomes a color strongly drawn by the color of clothes and the like, so that the light source estimation is easily erroneous.
From the above, it can be confirmed that the device is effective.

Further, the relationship between the multiple average value and the GWA application limit will be considered. The feature of the multiple average value is that high-saturation colors are excluded when the overall average of the image data is close to an achromatic color. Therefore, if a highly saturated color is on the screen and G
Even when WA does not hold, if the dominant color is an achromatic color, the overall average is close to an achromatic color, so that the light source estimation can be performed using the multiple average value. From this, the image is GWA
It can be said that there are cases where the light source can be estimated using the multiple average value even outside the applicable limit. Therefore, if the multiple average value is used, the conditions for GWA determination can be relaxed for an image in which a dominant color exists. Therefore, the image data determination accuracy is improved by increasing the threshold value of the vertical dispersion for determining whether or not the image in which the dominant color exists is within the GWA application limit. be able to. Furthermore, by estimating the light source using the multiple average value for an image in which a dominant color exists, the light source estimation accuracy can be improved.

On the other hand, according to the data analysis, an image in which a dominant achromatic color does not exist has a smaller GWA application limit range than an image in which a dominant achromatic color exists. Therefore, when it is determined whether an image in which a dominant color exists and an image in which a dominant color does not exist are within the GWA application limit using the same threshold value, an image in which a dominant color does not exist is determined. In some cases, a judgment error may occur. This problem distinguishes images with dominant colors from those that do not, and extends the GWA coverage for images with dominant colors.
For other images, the problem can be solved by narrowing the GWA application limit.

From the above description, it can be seen that even if the image is outside the GWA application limit, the light source estimation method using multiple average values is effective when the total average of the image data is close to achromatic. It became clear that there was. Therefore, it can be said that a necessary condition for the light source estimation method using the multiple average value to be effective is that the total average of the image data is close to achromatic. This means that the GWA is almost satisfied, and it is more appropriate to call it an extended GWA. However, for the sake of simplicity, the following description will simply call it GWA. At this time, considering light source estimation using the total average value or the multiple average value, it can be said that an image within the GWA application limit is an image suitable for light source estimation.

Here, a method of determining whether a dominant color exists on the screen will be considered. If there are two or more colors that have a large area on the screen, it cannot be said that there is a dominant color. In such a case, the amount of image data around the total average chromaticity and the chromaticity in the vicinity thereof on the chromaticity plane is reduced. On the other hand, when there is only one color occupying a large area on the screen, the amount of image data around the entire average chromaticity and the chromaticity in the vicinity thereof on the chromaticity plane increases. Therefore, whether or not a dominant color exists on the screen can be determined by examining the total average chromaticity and its vicinity as the reference chromaticity on the chromaticity plane, and examining the number of image data around the reference chromaticity. Can be determined.

When the number of image data around the multiple average chromaticity is sufficiently small, it indicates that the image data sampled from the screen is spread on the chromaticity plane, and is outside the GWA application limit. Can be determined. Since the multi-average chromaticity has a property of being close to the dominant color on the screen, it can be said that there is no dominant color when the number of image data around the multi-average chromaticity is small. From this, the image where the dominant color exists is GWA
Images that are within the application limits and otherwise are considered to be outside the GWA application limits. Therefore, an image in which a dominant color exists can be examined, and an image satisfying the condition can be determined to be within the GWA application limit, and further, an image suitable for light source estimation can be determined.

By the way, when the dominant color is not an achromatic color, the multiple average value near the dominant color does not represent the chromaticity of the light source. If the power is not turned on and the dominant color is not likely to be an achromatic color, the light source estimation result may be held.
However, when the power is turned on, the light source must be estimated. Therefore, when the dominant color is not an achromatic color, the light source can be correctly estimated by using the total average representing the center of gravity of the distribution of the sampled image data. As an example of this, there is a scene in which the sky and hay occupying a large area are simultaneously photographed outdoors. At this time, the total average chromaticity is in the middle between the chromaticity of the grass and the sky, but the multiple average value is close to the color of the grass and is shifted to the low color temperature side. Then, when using the multiple average value, the light source estimation is incorrect,
If the total average chromaticity is used, correct light source estimation can be performed.

As a method for determining whether or not the multiple average chromaticity is an achromatic color, a method for setting a specific area near the light source line shown in FIG. Conceivable. The above is the principle of the second invention.

Further, the principle of the third invention will be described. The color temperature of the white plate under the outdoor light source is high. On the other hand, the color temperature of the white plate under the fluorescent lamp is low. According to our analysis,
It is clear that when the sky enters the screen during outdoor shooting, the color of the sky has a high color temperature and high brightness. However, an image including a color having a high color temperature and a high luminance is not always taken outdoors. This is because the brightness of a bright blue image captured under a fluorescent lamp also increases at a high color temperature.

By the way, the blue color photographed under fluorescent light and the sky photographed outdoors have high brightness, but the absolute brightness of both is different. The reason why the brightness is high in both cases, although the absolute brightness is different, is that the state of aperture adjustment at the time of shooting is different. Generally, under a fluorescent light, the F-number representing the aperture state does not exceed a certain value unless the light source is directly projected. When a fluorescent lamp is directly projected, the F value increases, but the high-luminance portion has a low color temperature. Therefore, F
If the value is a certain value or more and a color with high luminance and high color temperature is included, it can be determined that the shooting environment at that time is outdoors. According to the analysis of the present inventor, colors having high luminance and high color temperature in an image photographed outdoors include not only the color of the sky but also the color of white clothes.

From the above, it can be said that a state including a color with high luminance and a high color temperature and a large F value appears characteristically only when the sky or the like is photographed outdoors. The outdoor light source detection unit of the present invention is a unit that detects such an image. For detection, a signal obtained by subjecting a control signal from the aperture adjustment mechanism control unit to a threshold value, and luminance data and chromaticity data obtained from image data are used. The above is the principle of the third invention.

(Embodiment 1) Next, a configuration of a first embodiment of the present invention will be described with reference to the drawings. FIG.
FIG. 1 is a block diagram showing a configuration of a light source estimation device of the present invention.

In FIG. 1, reference numeral 1 denotes a sampling circuit for converting an image signal IS from a CCD or the like in accordance with an appropriate sampling rule. Reference numeral 2 denotes an image signal IS sampled from the sampling circuit 1 as a luminance signal and two chromaticity signals. And 14 is a color chart recognizing unit that determines whether or not the subject is a color chart. FIG. 2 showing details of the color chart recognition unit.
, 3 is variance calculating means for calculating the variance of one chromaticity signal, 4 is variance calculating means for calculating the variance of the other chromaticity signal, and 5 is the covariance of the two chromaticity signals. And 6 a projection means for projecting a chromaticity signal in a direction perpendicular to the light source line described in the description of the principle and converting the chromaticity signal into a component size (ie, a vertical displacement from the light source line). Numeral 7 is a variance calculating means for receiving the vertical displacement from the projecting means 6 and calculating its variance.
Are variance calculation means 3, variance calculation means 4, and covariance calculation means 5
Is a quadratic equation solving means for receiving an output of each of the above and solving a quadratic equation having a coefficient composed of those values and outputting an appropriate solution. Reference numeral 9 denotes a color calculated by the chromaticity conversion unit 2. Coordinate conversion means for performing appropriate coordinate conversion on the degree coordinate value in accordance with the output of the quadratic equation solving means 8, and 10 is a correlation coefficient calculation means for calculating the correlation coefficient from the output of the coordinate conversion means 9. Where 11 and 12 are threshold value processing means for comparing the internally stored threshold value with the input signal and outputting a judgment signal, 13 is a normal AND circuit,
14 is a color chart recognition unit, 15 is a statistic calculation unit, and 16 is an image variance calculation unit.

In FIG. 1, reference numeral 17 denotes an average calculation means for receiving the chromaticity coordinate values from the luminance / chromaticity conversion unit 2 and calculating an average value thereof; It is an estimation unit. Reference numeral 19 denotes a luminance / chromaticity data storage unit that stores the luminance signal and the chromaticity signal converted by the luminance / chromaticity conversion unit 2 as luminance data and chromaticity data, respectively.
Reference numeral 20 denotes a storage control unit that prohibits storage in the luminance and chromaticity data storage unit until image determination is completed. 50 is for general images
And it consists of: in the light source chromaticity estimator, 2 1 Ri source chromaticity candidate calculation unit der for estimating a candidate of a light source chromaticity,
In FIG. 3A showing the details of the light source chromaticity candidate calculation unit, reference numeral 22 denotes an average calculation unit for calculating the total average of the chromaticity data, and reference numerals 23 and 24 denote input reference chromaticity and luminance chromaticity. A limited area data output unit for selecting and outputting data belonging to a limited area around the reference chromaticity from the chromaticity data read from the data storage unit;
An area limited average calculation unit 27 calculates the average of the limited area data output from the limited area data output unit. Reference numeral 27 denotes the total average chromaticity from the average calculation unit and the multiple average chromaticity from the area limited average calculation unit. Is a light source chromaticity candidate storage unit.

In FIG. 1, reference numeral 28 denotes a limited area center calculating section which receives and stores the multiple average chromaticity from the light source chromaticity candidate calculating section. Reference numeral 29 denotes a dominant color existence determining unit that determines whether or not a dominant color exists by checking the number of chromaticity data in the limited area, and the input reference chromaticity and the color read from the chromaticity data storage unit. A limited area data output unit 30 for selecting and outputting data belonging to a limited area around the reference chromaticity from the degree data, a counter 31 for counting the number of data output from the limited area data output unit, A dominant color presence determination threshold value processing unit 32 that determines that the dominant color is present when the output from the counter is greater than the set value, and determines that the dominant color is absent in other cases. A specific color determination unit 33 determines whether or not the light source chromaticity candidate is close to the light source line. Reference numeral 34 denotes a light source chromaticity determination unit that estimates the light source chromaticity from the light source chromaticity candidates, and gives priority to the total average chromaticity when the light source estimation control signal from the dominant color presence determination threshold value processing unit 32 is absent. A chromaticity, and if not, a priority chromaticity selection unit 35 that sets a multiple average chromaticity as a priority chromaticity. If not, a total chromaticity determination unit 36 is provided that uses the total average chromaticity as the estimated light source chromaticity.

Reference numeral 37 denotes a color chart recognition signal from the color chart recognition unit 14, a color chart estimated light source chromaticity signal from the color chart light source chromaticity estimation unit 18, and a general image estimated light source from the total chromaticity determination unit 36. An estimated light source chromaticity selection unit that receives the chromaticity signal and selects the estimated light source chromaticity according to whether the captured image is a color chart image or another general image. 38 receives the estimated illuminant chromaticity from the estimated illuminant chromaticity selection unit 37, appropriately divides the chromaticity plane, allocates the light source to each area, and estimates the light source based on which area the estimated illuminant chromaticity belongs to. This is a light source estimation unit. Reference numeral 39 denotes an aperture adjustment mechanism control unit that controls the aperture adjustment mechanism. Reference numeral 40 denotes an outdoor light source detection unit that determines whether or not a phenomenon that characteristically occurs when the sky or the like is photographed outdoors has occurred. FIG. 3 showing details of the outdoor light source detection unit 40.
In (b), reference numeral 41 denotes a threshold processing means for comparing a control signal from the aperture adjustment mechanism control unit with a threshold value held therein and outputting a determination signal; Threshold value processing means for reading out the luminance data from the section 19 and comparing it with a threshold value held therein to output a judgment signal; 43 reads out the chromaticity data from the luminance and chromaticity data storage section 19; A high color temperature color determination unit 44 for determining whether or not the chromaticity data belongs to a high color temperature area held therein; 44, an AND circuit; 45, an output from the AND circuit 44 inverted from 0 to 1 A detection control unit 46 counts the determination signal output from the high color temperature color determination unit 43 and outputs a control signal when all the chromaticity data have been determined. Is detected In response to a control signal from the control unit 46,
Threshold processing means for reading out the counted number from the counter 45, comparing the value with a threshold value held therein, and outputting a determination signal; 48, threshold processing means 41, 47; Is an AND circuit that receives a determination signal from the controller and outputs a comprehensive determination result as to whether or not the subject is illuminated with an outdoor light source.

In FIG. 1, reference numeral 49 denotes the outdoor light source detecting unit 4.
The light source determination unit receives the light source determination control signal from 0 and the estimated light source from the light source estimation unit 38, and comprehensively determines the light source.

The operation of the light source estimating apparatus configured as described above will be described. The input image signal IS will be described as (Y, RY, BY).
This is not binding on the invention. This is the same in the following embodiments, and will not be repeated.

First, an image signal IS from a CCD or the like enters the sampling circuit 1 and is converted according to an appropriate sampling rule. As an example, as shown in FIG. 5, the screen is divided into n × m blocks, and as shown in FIG. 6, a plurality of (k) data points are extracted from each block and averaged to obtain a block average <IS> i = Σ (Y _j , R−Y _j , B−Y _j ) / k where i = 1, 2,..., NN = n × mk determines the number of data sampled in each block. A method is conceivable. Brightness / chromaticity converter 2
Receives the image signal <IS> i = (YYi, RYYi, BYYi) sampled by the sampling circuit 1 and converts it into a chromaticity signal CSi = (BYi, RYi) where BYi = BYYi / YYi RYi = RYYi / YYi. Convert to In this embodiment, YYi is a luminance signal.

The color chart recognizing unit 14 receives the chromaticity coordinates obtained from the sampling data for one screen,
Whether or not it is the chromaticity distribution of the color chart image is recognized. Hereinafter, the operation of the color chart recognition unit 14 will be described with reference to FIG.

The variance calculation means 3 calculates the chromaticity coordinates BYi (i = 1, 2,..., N) obtained from the sampling data for one screen
Distributed VB = <(BYi- <BY> ) 2> However, <x>: calculating an average value? X _i / N of the variable x. The variance calculation means 4 calculates the chromaticity coordinates RYi (i = 1, 2,.
Distributed VR = <(RYi- <RY> ) 2> However, <x>: calculating an average value? X _i / N of the variable x. The covariance calculation means 5 calculates the chromaticity signal CSi = (BYi, R
From Yi), the covariance of BYi, RYi CV = <(BYi− <BY>) (RYi− <RY>)> where <x>: average value 変 数_xi / N of variable x.

The projecting means 6 receives the chromaticity signal CSi,
The direction perpendicular to the "light source line" described in the explanation of the principle (V
1, V2), and is converted into a component in that direction, that is, a vertical displacement from the “light source line”, Vi = V1 · BYi + V2 · RYi. Here, (V1, V2) is a two-dimensional unit vector. Distributed computing unit 7 receives the vertical displacement Vi from projection unit 6, the image signal variance VV = <(Vi- <V> ) 2> However, <x>: calculating an average value? X _i / N of the variable x I do. The quadratic equation solving means 8 includes the variance calculating means 3
Of the chromaticity signal CSi described in the explanation of the principle based on the variance VB calculated by The gradient, that is, the solution of the quadratic equation CV · a ² + (VB−VR) a−CV = 0 that satisfies 2CV · a + VB−VR ≧ 0 is calculated.

The coordinate transformation means 9 is a quadratic equation solving means 6
And the chromaticity signal CSi calculated by the chromaticity conversion unit 2 based on the gradient a of the “distribution axis” obtained in
(BYi, RYi) coordinate transformation on _{BY'i = A 11 · BYi + A} 12 · RYi RY'i = A subjected to _{21 · BYi + A 22 · RYi} , converted chromaticity coordinates (BY'i, RY′i) is calculated. Here, A ₁₁ , A ₁₂ , A ₂₁ , and A ₂₂ rotate the coordinate axes so that the “axis of distribution” is in the 45 ° direction.
The components of the × 2 matrix A are A ₁₁ = A ₂₂ = (1 + a) / [2 (1 + a ² )] ^1/2 −A ₁₂ = A ₂₁ = (1−a) / [2 (1 + a ² )] ^1/2 I have. The correlation coefficient calculation means 10 converts the converted chromaticity coordinates BY′i and RY′i obtained from the coordinate conversion means 9.
Absolute value of correlation coefficient RC = | Cov (BY ', RY') / [V (BY ') V (RY')] ^1/2 | where Cov (BY ', RY') = <(BYi- <BY>) (RYi-<RY>)> V (BY ') = <(BYi-<BY>) ² > V (RY') = <(RYi-<RY>) ² ><x>: variable x calculating the average value? x _i / N. This is nothing but the absolute value of the “rotational correlation coefficient” described in the explanation of the principle.

The threshold value processing means 11 compares the correlation coefficient RC calculated by the correlation coefficient calculation means 10 with the correlation coefficient threshold value θRC stored therein, and when RC <θRC, SC1 = 1: If there is no possibility of a color chart SC1 = 0: Not a color chart, and outputs a color chart correlation coefficient determination signal SC1. The threshold value processing means 12 compares the image signal variance VV calculated by the variance calculation means 7 with the image variance threshold value θVV stored therein. If VV> θVV, SC2 = 1: possible color chart If not, SC2 = 0: the color chart is not determined, and the color chart dispersion determination signal SC2 is output.

The AND circuit 13 includes a threshold processing unit 11
Of the color chart correlation coefficient determination signal SC1 output by the threshold value processing means 12 and the color chart dispersion determination signal SC2 output by the threshold value processing means 12
SC: SC = 1: Output color chart SC = 0: Output not color chart. Thus, the color chart is determined from the image signal.

The operation of the color chart recognition section 14 has been described above. The color chart light source chromaticity estimating unit 18 receives the chromaticity signal CSi from the luminance / chromaticity converting unit 2, calculates and outputs an average value thereof, and outputs it as a color chart estimated light source chromaticity signal JC.

The luminance / chromaticity data storage section 19 stores the chromaticity signal C
Si is stored as chromaticity data, and the luminance signal YYi is stored as luminance data. When the storage is completed, the storage control unit 20 prohibits the writing of the luminance data and the chromaticity data to the luminance and chromaticity data storage unit. Next, illuminance chromaticity estimation for general image
The setting unit 50 will be described.

The average calculation means 22 inside the light source chromaticity candidate calculation unit 21 reads out the chromaticity data from the luminance chromaticity data storage unit 19 and calculates the total average chromaticity. The limited area data output unit 23 receives the total average chromaticity signal from the average calculation unit 22 and sets it as the reference chromaticity, reads out the chromaticity data from the luminance chromaticity data storage unit 19, and reads the chromaticity data from the chromaticity data. Only data belonging to the restricted area around the output is output.
One example is the difference δCS between the chromaticity data and the reference chromaticity.
i is calculated, the inner product (δCSi, δCSi) is compared with a preset value θ, and if (δCSi, δCSi) <θ, it is determined that it exists in the area. A method of outputting data is conceivable. The area limited average calculation unit 25 averages the data output from the limited area data output unit 23 to obtain a double average chromaticity. The limited area data output unit 24 receives the double average chromaticity from the area limited average calculation unit 25 and sets it as the reference chromaticity, reads out the chromaticity data from the luminance chromaticity data storage unit 19, and, among the chromaticity data, Only data belonging to a restricted area around the reference chromaticity is output. The area limited average calculation unit 26 averages the data output from the limited area data output unit 24 and sets the average as triple average chromaticity. The light source chromaticity candidate storage unit 26 receives the total average chromaticity from the average calculation unit 22 and the triple average chromaticity from the area limited average calculation unit 26, and stores them as light source chromaticity candidates SS1 and SS2. . The limited area center calculation unit 28 receives the light source chromaticity candidate SS2 from the light source chromaticity candidate storage unit 27 and sets it as the limited area center chromaticity SSC. The limited area data output unit 30 inside the dominant color existence determination unit 29 is
From the chromaticity data storage unit 19, the chromaticity data is read out from the chromaticity data storage unit 19, and only the data belonging to the restricted area around the reference chromaticity is included in the chromaticity data. Is output. The counter 31 counts the number of data output from the limited area data output unit and sets the number as the limited area data number N3. Dominant color presence determination threshold processing unit 3
2 receives the limited area data number N3 and compares it with a threshold value θ1 held inside. When N3 ≧ θ1, J1 = 1 (presence of dominant color). When N3 <θ1, J1 = 0 (predominant color). Absence). The specific color determination unit 33 includes an estimated light source chromaticity storage unit 27.
From the light source chromaticity candidates SS1 and SS2, it is checked whether or not the light source chromaticity candidate SS2 enters a specific area on the chromaticity plane held therein. If yes, J2 = 1 (SS2 is a specific area) If not, it is assumed that J2 = 0 (SS2 is outside the specific area). The priority chromaticity selection unit 35 inside the light source chromaticity determination unit 34 receives the light source estimation control signal J1 from the dominant color presence determination threshold value processing unit 32, and when J1 = 1, J3 = 1 (3 When J1 = 0, J3 = 0 (overall average chromaticity SS1 priority). The overall chromaticity determination unit 36 includes a specific color determination signal J2 from the specific color determination unit 33, a priority chromaticity designation signal J3 from the priority chromaticity selection unit 35, and a light source chromaticity candidate SS1 from the light source chromaticity candidate storage unit 27. , SS2, when J2 = 1 and J3 = 1, SS = SS2 (adopting triple average chromaticity), otherwise, SS = SS1 (adopting total average chromaticity), and SS is estimated light source color Output as degrees.

The estimated light source chromaticity selection unit 37 includes a color chart recognition signal SC from the color chart recognition unit 14, a color chart estimated light source chromaticity signal JC from the color chart light source chromaticity estimation unit 18, and a light source chromaticity determination. Receiving the estimated light source chromaticity signal SS for general image from the unit 34, and by the value of the color chart recognition signal SC, SC = 1 (a color chart) → SJG = JC SC = 0 (a color chart) → Assuming that SJG = SS, an estimated light source chromaticity signal SJG is output.

The light source estimating unit 38 appropriately divides the chromaticity plane in advance and allocates light sources to the respective regions, receives the estimated light source chromaticity SJG from the estimated light source chromaticity selecting unit, The light source is estimated according to which region the SJG belongs to, and the light source estimation result is set to S1.

The aperture adjustment mechanism control section 39 receives the luminance signal and the chromaticity signal from the luminance / chromaticity conversion section 2 and calculates an aperture adjustment mechanism control signal CI for performing an optimal aperture adjustment. And outputs the aperture adjustment mechanism control signal CI to the threshold value processing means 41 inside the outdoor light source detection section 40. However, CI becomes maximum when the aperture is closed most. Therefore, as the F value increases, the CI also increases. The threshold value processing means 41 compares the aperture adjustment mechanism control signal CI with the internally held threshold value θCI. If CI ≧ θCI, JI1 = 1 (there is a possibility of an outdoor photographed image). JI1 = 0 (there is a possibility of an indoor photographed image). The threshold value processing means 42 reads out the luminance data YYi from the luminance / chromaticity data storage unit 19 and compares it with a threshold value θYY stored therein. When YYi ≧ θYY, JYi = 1 (high luminance color) If not, JYi = 0 (not a high brightness color). The high color temperature color determination section 43 reads the chromaticity data CSi from the luminance / chromaticity data storage section 19 and checks whether or not the chromaticity data CSi is within a preset area. Otherwise, JAi = 0 (not high color temperature color). The AND circuit 44 outputs 1 only when both JYi and JAi are 1, and outputs 0 otherwise. The counter 45 outputs 0 from the AND circuit 44.
Count the number of times of inversion from 1 to 1. The detection control unit 46 normally outputs 0, but counts the determination signals output from the high color temperature color determination unit 43, and outputs the control signal 1 when the determination for all chromaticity data is completed. The threshold processing unit 47 determines that the control signal output from the detection control unit 46 is 1
, The number NY counted from the counter 45
Is read, and NY is compared with a threshold value θNY held internally. If NY ≧ θNY, JI2 = 1 (there is a possibility of an outdoor photographed image). Otherwise, JI2 = 0 (the indoor photographed image There is a possibility). When the signal NY is output to the threshold value processing means 47, the counter value of the counter 45 is set to 0. AND
The circuit 48 receives the judgment signal JI from the threshold value processing means 41.
1 and the judgment signal JI2 from the threshold processing means 47, JI1 = JI2 = 0, JI3 = 1 (outdoor shot image), otherwise JI3 = 0 (indoor shot image possible ). The light source determination unit 49 is configured to control the light source determination control signal JI3 from the AND circuit 48
In response to 1, when JI3 = 1, Sf = OUT (outdoor light source) When JI3 = 0, Sf = S1.

In this way, the color chart and the general image other than the color chart are discriminated, and the image does not apply to any of the general images having the dominant color and the image having the characteristic characteristic of the image photographed outdoors. Determine the image and
A light source estimation device that estimates a light source by a method suitable for each is realized. Further, since the output from the color chart recognition unit 14 indicates whether or not the image is a color chart image, it can also serve as a color chart recognition device.

Further, since the output from the dominant color presence determination unit indicates whether or not the dominant color exists on the screen, it can also serve as a dominant color presence determination device.
Further, the output from the light source chromaticity determination unit 34 represents the estimation result of the light source chromaticity for the general image, and thus can also serve as a light source chromaticity estimating device.

The vector (V1, V2) for giving a direction perpendicular to the "light source line" need only be approximately parallel to the vector (1, 1), and need not be strictly determined. Although the accuracy of the color chart determination slightly varies depending on the parameters, the data analysis shows that the effect is small. The essence of the calculation of the image variance VV does not depend on whether (R, G, B) or (Y, RY, BY) is used as the image signal.

The image variance calculating section 16 calculates the image variance after converting the image signal into two-dimensional chromaticity coordinates. It is also possible to calculate the image variance directly from the three-dimensional image signal. In addition, there are many color chart determination conditions other than the method described in the present embodiment. For example,
It is also possible to prepare appropriate coefficients α and β, calculate CV = αRC−βVV, compare the CV with a threshold value, and output a color chart determination signal.

The “axis of distribution” is set to 4 by the coordinate conversion means 9.
The reason why the angle is set to the direction of 5 ° is that it is considered that it is most effective to adjust the angle to 45 °, and it is not always necessary to set the angle to 45 °. In addition, 2 used in the coordinate conversion means 7
Even if the coefficient 1 / [2 (1 + a ² )] ^1/2 common to the components of the × 2 matrix A is omitted or another one is used, the effect obtained last is the same.

Further, in this embodiment, the correlation coefficient is calculated again after the coordinate conversion. And the covariance before coordinate transformation,
It is not always necessary to perform coordinate transformation. By calculating the correlation coefficient after the coordinate conversion using these values without executing the coordinate conversion, a simple configuration with a smaller amount of calculation is possible.

Although the present embodiment uses the absolute value of the correlation coefficient, the same processing can be performed without using the absolute value. In the present embodiment, the “rotational correlation coefficient” of the chromaticity coordinates described in the explanation of the principle is used. However, the color chart can be recognized even by simply using the correlation coefficient of the chromaticity coordinates.

Further, in this embodiment, the limited area center chromaticity SSC is a triple average chromaticity, but may be a total average chromaticity or a double average chromaticity. The double average may be used as the light source chromaticity candidate SS1 instead of the total average. The double average is the average of the image data in a limited area around the entire average on the chromaticity plane, so when there is hay and the sky on the screen, the triple average value is closer to the color of the hay. It does not become.

In the dominant color existence determination unit, whether or not a dominant color exists is determined by using the number of data in the limited area. It may be determined that the color exists, and otherwise, it may be determined that the dominant color is absent.

In this embodiment, it is determined whether or not a dominant color exists on the screen by using image data around the triple average chromaticity. However, the number of data used when calculating the triple average chromaticity is determined. Can be used to determine whether a dominant color exists. At this time, the configuration of the dominant color presence determination unit can be simplified.

Further, the limited area data output unit outputs chromaticity data such that the distance from the reference chromaticity is equal to or less than the set value θ. The degree data may be output. At this time, if a plurality of points that determine the polygon are held in advance, there is no problem in configuration.

In this embodiment, when the judgment result by the outdoor light source detecting section is “the subject may be an indoor image”, the light source estimation based on the color chart recognition result, the dominant color existence judgment result, and the GWA is performed. Although the result is output, the light source estimation result to be output at this time has no problem in the configuration even if the light source is estimated by any method. Although the detection control unit counts the output from the high color temperature color judgment unit, it counts the output from the threshold processing unit 42 that performs threshold processing on the luminance data, and calculates the luminance data and the chromaticity data. May be configured to determine whether the processing for is completed. These matters are the same in the following embodiments and will not be repeated.

(Embodiment 2) Next, the configuration of a second embodiment of the present invention will be described with reference to the drawings. FIG.
FIG. 3 is a block diagram showing a configuration of a color chart recognition device.

In FIG. 9, reference numeral 1 denotes a sampling circuit for converting an image signal IS from a CCD or the like in accordance with an appropriate sampling rule, and 100 converts the image signal IS sampled from the sampling circuit 1 into two chromaticity signals. A chromaticity conversion unit 71 is a projecting means for projecting a chromaticity signal in a direction perpendicular to the light source line described in the explanation of the principle and converting the chromaticity signal into the size of the component (ie, vertical displacement from the light source line). Reference numeral 72 denotes projection means for projecting a chromaticity signal in a direction parallel to the light source line and converting the chromaticity signal into a component size (that is, displacement in a parallel direction along the light source line). Reference numeral 3 denotes a vertical direction from the projection means 71. A variance calculating means for calculating the variance in response to the displacement, 4 is a variance calculating means for calculating the variance in response to the parallel displacement from the projecting means 72, and 5 is a variance calculating means. Receiving a direction parallel displacement from vertical displacement and projection means 72 from the means 71 is a covariance calculation means for calculating their covariance, 8 distributed computing unit 3
And quadratic equation solving means for receiving an output of each of the variance calculating means 4 and the covariance calculating means 5, solving a quadratic equation having coefficients constituted by the values, and outputting an appropriate solution. Is an arithmetic circuit that receives the outputs of the variance calculation means 3, the variance calculation means 4, the covariance calculation means 5, and the quadratic equation solving means 8 and calculates and outputs a rotational correlation coefficient using these outputs; Reference numerals 11 and 12 denote threshold value processing means for comparing the threshold value stored therein with the input signal and outputting a judgment signal. Reference numeral 13 denotes an AND circuit.
Denotes a color chart determination unit which is the core of the present invention, 102 denotes a statistic calculation unit, and 103 denotes an image variance calculation unit.

The operation of the color chart recognizing device thus constructed will be described with reference to FIG. First, an image signal IS from a CCD or the like enters the sampling circuit 1 and is converted according to an appropriate sampling rule. The chromaticity converter 100 converts the image signal <IS> i = (YYi, RYY) sampled by the sampling circuit 1
i, BYYi), and receives the chromaticity signal CSi = (B
Yi, RYi).

The projection means 71 receives the chromaticity signal CSi, and receives the chromaticity signal CSi in a direction perpendicular to the "light source line" described in the description of the principle.
This is projected onto (V1, V2), and is converted into a component in that direction, that is, a vertical displacement Vi = V1.BYi + V2.RYi from the "light source line". Here, (V1, V2) is a two-dimensional unit vector. The projection means 72 outputs the chromaticity signal CSi
Accordingly, the light is projected in a direction (V2, −V1) parallel to the “light source line” described in the description of the principle, and a component in that direction, that is, a parallel direction displacement from the “light source line” Ui = V2 · BYi− V1 · RYi. Distributed computing unit 3 receives the vertical displacement Vi from projection unit 71, the vertical direction image signal variance VV = <(Vi- <V> ) 2> However, <x>: the average value of the variable x? X _i / N Is calculated. Distributed computing unit 4 receives the parallel displacement Ui from the projection means 72, parallel image signal variance VU = <(Ui- <U> ) 2> However, <x>: average value? X _i / N of the variable x Is calculated.

The covariance calculation means 5 calculates the vertical displacement Vi from the projection means 71 and the parallel displacement U from the projection means 72.
In response to i, these covariances CVU = <(Vi- <V>) (Ui- <U>)> where <x>: average value 変 数_xi / N of variable x. The quadratic equation solving means 8 includes the variance calculating means 3
Receiving the variance VV calculated in the above, the variance VU calculated by the variance calculation means 4 and the covariance CVU calculated by the covariance calculation means 5,
The gradient of the “distribution axis” of the chromaticity signal CSi with respect to the direction perpendicular to the “light source line”, that is, 2CVU · a + VV of the solution of the quadratic equation CVU · a ² + (VV−VU) a−CVU = 0 -Calculate those that satisfy VU ≥ 0.

The arithmetic circuit 73 calculates the gradient a of the “distribution axis” obtained by the quadratic equation solving means 8, the variance VV calculated by the variance calculating means 3, the variance VU calculated by the variance calculating means 4, and the covariance calculation. Given the covariance CVU calculated by the means 5, the displacement X
i = (Vi, Ui) transformed displacement X when subjected to coordinate transformation _{V'i = A 11 · Vi + A} 12 · Ui U'i = A 21 · Vi + A 22 · Ui against '= (V'i, U'i)
Calculate the absolute value of the correlation coefficient of Where A ₁₁ , A ₁₂ , A ₂₁ ,
A ₂₂ is a component of the 2 × 2 matrix A that rotates the coordinate axes so that the “axis of distribution” is in the 45 ° direction. A ₁₁ = A ₂₂ = (1 + a) / [2 (1 + a ² )] ^1/2 −A _{12 = a 21 = (1-a} ) / [2 (1 + a 2)] is set to ^1/2. Here, the absolute value of the correlation coefficient after the coordinate transformation can be calculated from the variance and the covariance before the coordinate transformation using the following equation.

RC ′ = | Cov (v ′, u ′) / [V (v ′) V (u ′)] ^1/2 | where Cov (v ′, u ′) = A ₁₁ A ₂₁ VV + (A _{11 a 22 + a 12 a 21)} CVU + a 12 a 22 VU V (v ') = a 11 2 VV + a 11 a 12 CVU + a 12 2 VU V (u') = a 21 2 VV + a 21 a 22 CVU + a 22 2 VU This principle This is nothing less than the absolute value of the “rotational correlation coefficient” described in the description.

The threshold value processing means 11 compares the rotational correlation coefficient RC ′ calculated by the arithmetic circuit 73 with the correlation coefficient threshold value θRC ′ stored therein, and if RC ′ <θRC ′ SC1 = 1: If there is no possibility of a color chart SC1 = 0: Not a color chart, and outputs a color chart correlation coefficient determination signal SC1. The threshold value processing unit 12 compares the image signal variance VV calculated by the variance calculation unit 3 with the image variance threshold value θVV stored therein. If VV> θVV, SC2 = 1: color chart possible If not, SC2 = 0: It is determined that the color chart is not displayed, and the color chart dispersion determination signal SC2 is output.

The AND circuit 13 includes a threshold processing unit 11
Of the color chart correlation coefficient determination signal SC1 output by the threshold value processing means 12 and the color chart dispersion determination signal SC2 output by the threshold value processing means 12
SC: SC = 1: Output color chart SC = 0: Output not color chart. Thus, the color chart is determined from the image signal.

The feature of this embodiment is that the color chart recognizing unit of the first embodiment requires three variance calculations, whereas the coordinate axes of the chromaticity data are reset to the direction perpendicular to the “light source line” and the direction parallel thereto. By calculating the image variance, only two variance calculations are required, and the amount of calculation is smaller and the configuration is simpler.

In this embodiment, when calculating the “rotational correlation coefficient”, instead of performing coordinate transformation, the variance and covariance before the first-order transformation and the variance and covariance after the first-order transformation are generally calculated. Although the amount of calculation is reduced and the configuration is simplified by using the relationship that is established, such a configuration is not necessarily required, and coordinate conversion is performed as in the color chart recognition unit of the first embodiment. It may be configured.

(Embodiment 3) Next, a third embodiment of the present invention will be described with reference to the drawings. FIG.
It is a block diagram showing the composition of the image data judging device provided with the dominant color existence judging part which is an Example of.

In FIG. 11, reference numeral 1 denotes a sampling circuit for converting an image signal IS from a CCD or the like in accordance with an appropriate sampling rule; 100, a chromaticity conversion unit for converting sampled image data into a chromaticity signal; A chromaticity data storage unit that stores the obtained chromaticity signal as chromaticity data, and a storage control unit 20 that prohibits storage in the chromaticity data storage unit until image determination is completed. Reference numeral 201 denotes a limited area center calculation unit that calculates the chromaticity at the center of the limited area, an average calculation unit 22 that calculates the total average chromaticity of the chromaticity data, and an input reference chromaticity and chromaticity data storage unit. A limited area data output unit 23 that selects and outputs data belonging to a limited area around the reference chromaticity from the chromaticity data read from the limited chromaticity data; And an area-limited average calculating unit 25 for calculating an average. Reference numeral 202 denotes a dominant color existence determination unit that determines whether or not a dominant color exists on the screen, and determines a reference chromaticity based on the input reference chromaticity and chromaticity data read from the chromaticity data storage unit. Limited area data output unit 3 for selecting and outputting data belonging to a limited area around it
0, a counter 31 for counting the number of data output from the limited area data output unit, and when the output from the counter 31 is larger than the first set value, it is determined that there is a dominant color, and the control color is larger than the second set value. A dominant color presence determination threshold value processing unit 203 that determines that the GWA is not established when the value is smaller and that the dominant color is absent in other cases.

A determination result processing unit 204 processes the determination result from the dominant color presence determination unit 202 and checks whether GWA is established or not to determine whether the image is suitable for light source estimation. Projecting means 205 for projecting in a particular direction and converting the displacement into a displacement which is the size of the component in that direction; dispersion calculating means 206 for calculating the variance of the displacement obtained from the chromaticity data; In response to the threshold control signal from unit 203, a large threshold is set when a dominant color is present, a small threshold is set when a dominant color is absent, and a threshold is set when GWA is not established. Receiving a variance from the variance calculation means and a threshold from the threshold setting unit,
A GWA determination threshold value processing unit 208 determines that the GWA is established when the variance is smaller than the threshold value, and otherwise determines that the GWA is not established.

The operation of the image data judging device configured as described above will be described. First, an image signal IS = (Y, RY, BY) from a CCD or the like enters the sampling circuit 1 and is converted according to an appropriate sampling rule. The chromaticity conversion unit 100 receives the block average (YYi, RYYi, BYYi) from the sampling circuit 1, calculates BYi = BYYi / YYi, RYi = RYYi / YYi, and converts it into a chromaticity signal CSi = (BYi, RYi). The chromaticity data storage unit 200 stores the chromaticity signal CSi as chromaticity data. When the storage is completed, the storage control unit 20 prohibits the writing of the chromaticity data to the chromaticity data storage unit.

The average calculating means 22 inside the limited area center calculating section 201 reads out the chromaticity data from the chromaticity data storage section 200 and calculates the total average chromaticity. The limited area data output unit 23 receives the total average chromaticity signal from the average calculation unit 22 and sets it as the reference chromaticity, reads out the chromaticity data from the chromaticity data storage unit 200, and sets the reference chromaticity Only the data belonging to the restricted area around is output. The area limited average calculation unit 25 averages the data output from the limited area data output unit 23 to obtain a double average chromaticity. The limited area data output unit 30 inside the dominant color existence determination unit 202 receives the double average chromaticity from the area limited average calculation unit 25 and sets it as the reference chromaticity, and reads the chromaticity data from the chromaticity data storage unit 200. However, of the chromaticity data, only data belonging to a restricted area around the reference chromaticity is output. The counter 31 counts the number of data output from the limited area data output unit and sets the number as limited area data number N2. The achromatic color existence determination threshold value processing unit 203 calculates the limited area data number N
And N2 ≧ 2, J1 = 1 (presence of dominant color), θ1> N2 ≧ θ2, J1 = 0 (absence of predominant color) N2 When <θ2, J1 = 2 (GWA not established). Projection means 2 inside judgment result processing section 204
05 is the chromaticity data CSi from the chromaticity data storage unit 200.
= (C1i, C2i), i = 1,..., N, and project in the direction (V1, V2) perpendicular to the light source line in FIG.
The magnitude of the component in that direction, that is, the displacement from the light source line Xi = BYi · V1 + RYi · V2 <X> = (ΣXi) / N The variance calculation unit 206 receives the displacement Xi, <X> from the projection unit 205 and calculates the variance SGM = {(Xi− <X>) ² } / N. The threshold value setting unit 207 internally stores a set value θ
3 and θ4, and receives the threshold value control signal J1 from the dominant color presence determination threshold value processing unit 203. When J1 = 1, θs = θ3, when J1 = 0, θs = θ4 J1 = 2, θs = 0. However, θ3 ≧ θ4. The GWA determination threshold value processing unit 208 receives the image data variance SGM from the variance calculation unit 206 and the threshold value signal θs from the threshold value setting unit 207, and when SGM <θs, J4 = 1 (GWA holds) In other cases, J4 = 0 (GWA not established).

In this way, an image in which a dominant color exists and an image in which it does not exist are discriminated, and a threshold value for GWA judgment is set using the discrimination result, and GWA is determined from the magnitude relationship between the vertical dispersion and the set threshold value. By determining whether or not is established, an image data determination device that determines an image suitable for light source estimation is realized.

In the present embodiment, the threshold control signal J1 output from the dominant color existence determining section is three, that is, dominant color exists, dominant color does not exist, and GWA does not hold. When the number of chromaticity data N2 is sufficiently small, the vertical variance becomes large and the GWA determination unit often determines that GWA is not established. Therefore, J1 may be two types, that is, the presence of a dominant color and the absence of a dominant color. If the determination result from the dominant color presence determination unit is that there is a dominant color,
The determination result processing unit may be configured so that the determination is made, and otherwise, the GWA is not determined.

Furthermore, the threshold control signal J1 output from the dominant color existence determination section may be of three or more types. At this time, by increasing the set value held inside the threshold value setting unit, the threshold value θs for GWA determination can be set more finely, and the image data determination accuracy is improved.
Since these matters are the same in the following examples,
Do not repeat.

(Embodiment 4) Next, a fourth embodiment of the present invention will be described with reference to the drawings. The fourth embodiment is a light source estimating device including a dominant color existence determination unit and a specific color determination unit. Although the light source chromaticity estimating unit has been described in the first embodiment, and the image data determining device has been described in the third embodiment, when considered as a part of the light source estimating device, the dominant color existence determining unit is a common part, One of the light source chromaticity candidates can be the limited area center chromaticity. When the dominant color presence determination unit is used as a part of the light source estimating unit, the configuration of the dominant color presence determination unit so as to output a detailed determination result improves the light source estimation accuracy. Therefore, a fourth embodiment describes a light source estimating apparatus that includes a light source chromaticity estimation portion and an image data determination portion, and that performs light source estimation using the estimated light source chromaticity and the determination result.

FIG. 12 is a block diagram showing the configuration of a light source estimation device according to the fourth embodiment. In FIG. 12, 1
Is a sampling circuit for converting an image signal IS from a CCD or the like according to an appropriate sampling rule, 100 is a chromaticity conversion unit for converting sampled image data into a chromaticity signal, and 200 is a chromaticity signal for converting the converted chromaticity signal. A chromaticity data storage unit 20 that stores data is a storage control unit that prohibits storage in the chromaticity data storage unit until image determination is completed. Reference numeral 301 denotes a light source chromaticity estimating unit for estimating the light source chromaticity, which is read from the average calculating unit 22 for calculating the total average chromaticity of the chromaticity data, and the input reference chromaticity and chromaticity data storage unit. Limited area data output units 23 and 24 for selecting and outputting data belonging to a limited area around the reference chromaticity from the chromaticity data, and calculating an average of limited area data output from the limited area data output unit Region limited average calculation units 25 and 26, a light source chromaticity candidate storage unit 27 that stores the total average chromaticity from the average calculation unit and the multiple average chromaticity from the region limited average calculation unit, When the color of the light source from body radiation is the parameter of the temperature of the black body as a parameter, the light source chromaticity candidate is determined to be a specific color if the light source chromaticity candidate is close to a straight line that approximates the locus drawn on the chromaticity plane, and otherwise specified. Specific color determination unit 33 that determines that the color is not a color , Dominant color absence or GWA
Priority chromaticity selection unit 3 that sets the overall average chromaticity as the priority chromaticity if the condition is not satisfied, and sets the multiple average chromaticity as the priority chromaticity otherwise.
02, when the total average chromaticity is the priority chromaticity, the total average chromaticity is the estimated light source chromaticity, the multiple average chromaticity is the priority chromaticity, and
If the specific color determination signal from the specific color determination unit is a specific color, the multi-average chromaticity is set as the estimated light source chromaticity, and if not applicable in any case, the total chromaticity is set to "conditionally total average chromaticity" 36.

Reference numeral 303 denotes a dominant color presence determination unit, which includes a counter 31 for counting the number of data output from the limited area data output unit 24, and a first output from the counter.
Is set to be larger than the second set value, and GWA is set when the output from the counter is smaller than the second set value.
If the output from the counter is greater than the second set value and smaller than the third set value, the control color is determined to be absent, and otherwise the control color existence possibility is determined to be small, and the control color presence determination threshold is determined. A value processing unit 304. 305
Is a determination result processing unit that processes the determination result from the dominant color presence determination unit 303 and checks whether GWA is established to determine whether the image is suitable for light source estimation. When the color of the light source from the camera as a parameter is the temperature of the black body, the projection unit 205 projects the trajectory drawn on the chromaticity plane in a direction perpendicular to a straight line approximating the trajectory and converts the trajectory into a displacement which is the size of the component in that direction. And a variance calculating means 206 for calculating the variance of the displacement obtained from the chromaticity data, and a threshold control signal from the dominant color presence determination threshold value processing unit 304 to determine whether or not there is a dominant color. Set the threshold,
A threshold setting unit 30 that sets a threshold to 0 when GWA is not established, and sets a small threshold in other cases.
6 and the variance from the variance calculation means and the threshold value from the threshold value setting unit, and if the variance is smaller than the threshold value, the GWA is determined to be established; otherwise, the GWA is determined to be not established. The threshold value processing unit 208 and the estimated illuminant chromaticity determined by the overall chromaticity determination unit are appropriately divided on the chromaticity plane, and the light sources are allocated to the respective regions. Light source estimating section 307 for estimating
A comprehensive determination unit 308 that performs a comprehensive determination using the light source estimation control signal from the A determination threshold value processing unit and the light source estimation result from the light source estimation unit, and receives the comprehensive determination result from the comprehensive determination unit. And a light source estimation result holding unit 309 that outputs the light source estimation result held therein when the determination of the light source cannot be made.

The operation of the light source estimating apparatus configured as described above will be described. First, an image signal IS = (Y, RY, BY) from a CCD or the like enters the sampling circuit 1 and is converted according to an appropriate sampling rule. The chromaticity conversion unit 100 receives the block average (YYi, RYYi, BYYi) from the sampling circuit 1, calculates BYi = BYYi / YYi, RYi = RYYi / YYi, and converts it into a chromaticity signal CSi = (BYi, RYi). The chromaticity data storage unit 200 stores the chromaticity signal CSi as chromaticity data. When the storage is completed, the storage control unit 20 prohibits the writing of the chromaticity data to the chromaticity data storage unit.

The average calculating means 22 inside the light source chromaticity estimating unit 301 reads out the chromaticity data from the chromaticity data storage unit 200 and calculates the total average chromaticity. The limited area data output unit 23 receives the total average chromaticity signal from the average calculation unit 22 and sets it as the reference chromaticity, reads out the chromaticity data from the chromaticity data storage unit 200, and sets the reference chromaticity Only the data belonging to the restricted area around is output. The area limited average calculation unit 25 averages the data output from the limited area data output unit 23 to obtain a double average chromaticity. The limited area data output unit 24 outputs the limited area average calculation unit 25
The chromaticity data storage unit 2 receives the weighted average chromaticity and sets it as a reference chromaticity.
The chromaticity data is read from 00, and only the data belonging to the restricted area around the reference chromaticity is output from the chromaticity data. The area limited average calculation unit 26 averages the data output from the limited area data output unit 24 and sets the average as triple average chromaticity. The light source chromaticity candidate storage unit 27 receives the total average chromaticity from the average calculation unit and the triple average chromaticity from the area limited average calculation unit 26, and stores them in the light source chromaticity candidates SS1, SS
Stored as 2.

The counter 31 inside the dominant color existence determining section 303 counts the number of data output from the limited area data output section 24 and sets the number as limited area data number N2. The dominant color presence determination threshold value processing unit 304 receives the limited area data number N2, and stores the threshold values θ1, θ2,
Compared with θ1 ′, when N2 ≧ θ1, J1 = 1 (presence of dominant color) θ1> N2 ≧ θ1 ′, J1 = 3 (small possibility of dominant color existence) θ1 ′> N2 ≧ θ2, J1 = 0 (absence of dominant color) When N2 <θ2, J1 = 2 (GWA not established).

The specific color determination unit 33 receives the light source chromaticity candidates SS1 and SS2 from the light source chromaticity candidate storage unit 27, and determines whether or not the light source chromaticity candidate SS2 enters a specific area on the chromaticity plane held therein. Investigation, if it is included, J2 = 1 (SS2 is inside the specific area) If not, set J2 = 0 (SS2 is outside the specific area). The priority chromaticity selection unit 302 receives the light source estimation control signal J1 from the dominant color existence determination threshold value processing unit 304, and when J1 = 1 or J1 = 3, J3 = 1 (triple average chromaticity SS2 priority) When J1 = 0 or J1 = 2, J3 = 0 (the overall average chromaticity SS1 has priority). The overall chromaticity determination unit 36 includes a specific color determination signal J2 from the specific color determination unit, a priority chromaticity designation signal J3 from the priority chromaticity selection unit, and light source chromaticity candidates SS1 and SS2 from the light source chromaticity candidate storage unit 27. When J2 = 1 and J3 = 1, SS = SS2 (adopting triple average chromaticity) SJJ = 0 (no condition) When J2 = 0, SS = SS1 (adopting total average chromaticity) SJJ = 0 (no condition) When J2 = 1 and J3 = 0, SS = SS1 (total chromaticity is adopted) SJJ = 1 (basically undeterminable), and (SS, SJJ) is estimated light source chromaticity Output as

The projection means 205 inside the determination result processing unit 305 stores the chromaticity data C
In response to Si = (BYi, RYi), i = 1,..., N, the light is projected in the direction (V1, V2) perpendicular to the light source line in FIG. Xi = BYi · V1 + RYi · V2 <X> = (ΣXi) / N The variance calculation unit 206 receives the displacement Xi, <X> from the projection unit 205 and calculates the variance SGM = {(Xi− <X>) ² } / N. The threshold value setting unit 306 internally stores the set value θ
3 and θ4, and receives the threshold control signal J1 from the dominant color presence determination threshold value processing unit 304. When J1 = 1, θs = θ3 When J1 = 0 or J1 = 3, θs = θ4 When J1 = 2, θs = 0. However, θ3 ≧ θ4. The GWA determination threshold value processing unit 208 receives the image data variance SGM from the variance calculation unit 206 and the threshold value signal θs from the threshold value setting unit 207, and when SGM <θs, J4 = 1 (GWA holds) In other cases, J4 = 0 (GWA not established).

The illuminant estimating unit 307 appropriately divides the chromaticity plane in advance and allocates the illuminant to each area, receives the estimated illuminant chromaticity SS from the overall chromaticity determining unit, and receives the estimated illuminant chromaticity SS. The light source is estimated based on which region the light source belongs to, and the light source estimation result is set to S1. Then, it outputs the light source estimation result S1 and the condition signal SJJ from the overall chromaticity determination unit.

The overall judgment unit 308 is provided with the light source estimation unit 307 and G
The primary light source estimation result S1, the condition signal SSJ, and the light source estimation control signal J
4, when rising, S2 = S1 when non-rising J4 = 1 (GWA established) and SSJ =
When 0, S2 = S1 When non-rising, J4 = 0 (GWA not established) or SS
When J = 1, it is determined that S2 = impossible, and the secondary light source estimation result S2 is output. The light source estimation result holding unit 309 receives the secondary light source estimation result S2 from the comprehensive judgment unit 308, and when S2 ≠ cannot be determined Sf = S2 Subf = S2 S2 = when determination is not possible Sf = Sbuf Assuming that 2 The next light source estimation result is held in Sbuf, and the final light source estimation result Sf is output.

In this manner, an image having a dominant color is discriminated from an image not having a dominant color, and a threshold value for GWA determination is set by using the discriminated result, and the GWA is determined based on the magnitude relationship between the vertical dispersion and the set threshold value. Is determined, an image suitable for light source estimation is determined, and a light source estimation device that estimates a light source using only an image suitable for light source estimation is realized.

A luminance / chromaticity conversion unit for converting the sampled image signal IS into a luminance signal and two chromaticity signals, and a luminance signal and a chromaticity signal converted by the luminance / chromaticity conversion unit are respectively represented by luminance data, A luminance and chromaticity data storage unit that stores as chromaticity data, a storage control unit that prohibits storage in the luminance and chromaticity data storage unit until the light source estimation ends, an aperture adjustment mechanism control unit that controls an aperture adjustment mechanism, An outdoor light source detection unit that determines whether a phenomenon that appears characteristically when shooting the sky outdoors, etc., and a light source estimation result from the light source result holding unit and a detection result from the outdoor light source detection unit are comprehensively received. By adding a light source determination unit for determining the light source to the light source, it is possible to improve the light source estimation accuracy. The operations of the luminance / chromaticity converter, the luminance / chromaticity data storage, the aperture adjustment mechanism controller, and the outdoor light source detector are the same as those in the first embodiment. The light source determination unit sets the light source estimation result to the outdoor light source when the characteristic outdoor captured image can be detected by the outdoor light source detection unit; otherwise, the light source estimation result output from the light source estimation result holding unit Is output. In this way, a light source estimating apparatus capable of correctly estimating the light source when shooting the sky or the like outdoors is realized.

[0123]

As described above, according to the present invention, an image signal
Circuit that converts according to simple sampling rules
And the sampled N pieces (N is an integer of 2 or more)
Image signal is converted to N chromaticities as coordinates on the chromaticity plane
Chromaticity conversion unit, and the average coordinates of the N chromaticities are all averaged.
Calculated as chromaticity, and among the N chromaticities, the total average chromaticity
The average coordinates of chromaticity existing in the limited area of
The total average chromaticity and multiple average chromaticity are determined as light source chromaticity candidates.
A light source chromaticity candidate calculating unit to perform
Total number of chromaticities included in the limited area centered on the weighted average chromaticity
There is a dominant color that indicates a color that is often included in the image from
A dominant color presence determining unit for determining whether the dominant color is present
One of the light source chromaticity candidates is determined from the determination result from the
A light source chromaticity determination unit that selects and sets the light source chromaticity, and the light source chromaticity
And a light source estimating unit for estimating the light source to which region on the chromaticity plane belongs to
The correct color tone in video cameras, etc.
Can be reproduced.

[0124]

[0125]

[0126]

[0127]

[0128]

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a light source estimation device according to a first embodiment of the present invention.

FIG. 2 is a block connection diagram showing a configuration of a color chart recognition unit in the embodiment.

FIG. 3A is a block connection diagram illustrating a configuration of a light source chromaticity candidate calculation unit according to the embodiment; and FIG. 3B is a block connection diagram illustrating a configuration of an outdoor light source detection unit according to the embodiment.

FIG. 4 is a conceptual diagram showing a “light source line” and a specific area near the “light source line” in a color space for explaining the principle of the present invention.

FIG. 5 is a conceptual diagram showing an example of a sampling format of image data.

FIG. 6 is a conceptual diagram showing an example of a sampling format of image data.

FIG. 7 is a conceptual diagram of a distribution between a color chart image and a general image on a variance- “rotational correlation coefficient” plane for explaining the principle of the present invention.

FIG. 8 is a flowchart showing the operation of a color chart recognition unit according to the first embodiment of the present invention.

FIG. 9 is a block diagram showing a configuration of a color chart recognition device according to a second embodiment of the present invention.

FIG. 10 is a flowchart showing the operation of the color chart recognition device in the embodiment.

FIG. 11 is a block diagram illustrating a configuration of an image data determination device according to a third embodiment of the present invention.

FIG. 12 is a block diagram illustrating a configuration of a light source estimation device according to a fourth embodiment of the present invention.

[Explanation of symbols]

REFERENCE SIGNS LIST 1 sampling circuit 2 luminance and chromaticity conversion unit 3 variance calculation means 4 variance calculation means 5 covariance calculation means 6 projection means 7 variance calculation means 8 quadratic equation solving means 9 coordinate conversion means 10 correlation coefficient calculation means 11 threshold processing Means 12 Threshold processing means 13 AND circuit 14 Color chart recognition unit 15 Statistics calculation unit 16 Image variance calculation unit 17 Average calculation means 18 Color chart light source chromaticity estimation unit 19 Luminance chromaticity data storage unit 20 Storage control unit 21 Light source chromaticity candidate calculation unit 22 Average calculation means 23 Limited area data output unit 24 Limited area data output unit 25 Area limited average calculation unit 26 Area limited average calculation unit 27 Light source chromaticity candidate storage unit 28 Limited area center calculation unit 29 Dominant color Presence determination unit 30 Limited area data output unit 31 Counter 32 Dominant color presence determination threshold processing unit 33 Specific color determination unit 34 Light source chromaticity determination unit 35 Priority chromaticity selection unit 36 Overall chromaticity determination unit 37 Estimated light source chromaticity selection unit 38 Light source estimation unit 39 Aperture adjustment mechanism control unit 40 Threshold processing unit 41 Threshold processing Means 42 Threshold processing means 43 High color temperature determination unit 44 AND circuit 45 Counter 46 Detection control unit 47 Threshold processing means 48 AND circuit 49 Light source determination unit 50 Light source chromaticity estimation unit for general image 71 Projection unit 72 Projection unit 73 arithmetic circuit 100 chromaticity conversion unit 101 color chart recognition unit 102 statistic calculation unit 103 image variance calculation unit 200 chromaticity data storage unit 201 limited area center calculation unit 202 dominant color existence determination unit 203 dominant color existence determination threshold processing Unit 204 Judgment result processing unit 205 Projection unit 206 Variance calculation unit 207 Threshold setting unit 208 GWA Constant threshold processing section 301 Light source chromaticity estimating section 302 Priority chromaticity selecting section 303 Dominant color presence determining section 304 Dominant color presence determining threshold processing section 305 Decision result processing section 306 Threshold setting section 307 Light source estimating section 308 Comprehensive judgment unit 309 Light source estimation result holding unit

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-5-49048 (JP, A) JP-A-5-49049 (JP, A) JP-A-5-111059 (JP, A) JP-A-63- 13489 (JP, A) JP-A-3-16494 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H04N 17/02 H04N 9/73

Claims (6)

(57) [Claims] 1. A sampling circuit for converting an image signal in accordance with the appropriate sampling rule, the sampling of
N (N is an integer of 2 or more) image signals on the chromaticity plane
A chromaticity conversion unit for converting into N chromaticities as coordinates of
The average coordinates of N chromaticities are determined as the total average chromaticity,
Colors present in the limited area of the total average chromaticity among the chromaticities
The average coordinates of degrees are determined as multiple average chromaticities,
Light source chromaticity candidate calculator that uses multiple average chromaticity as a light source chromaticity candidate
And centering on the multiple average chromaticity of the N chromaticities
From the total number of chromaticities included in the limited area, many are included in the image
Dominant color exists to determine if there is a dominant color indicating the color
From the determination result from the determination unit and the dominant color presence determination unit
Light that selects one of the light source chromaticity candidates and sets it as the light source chromaticity
A light source estimating apparatus comprising: a source chromaticity determining unit; and a light source estimating unit that estimates a light source based on which region on the chromaticity plane the light source chromaticity belongs to. 2. A light source chromaticity candidate calculation unit , comprising:
Means for calculating average coordinates as the total average chromaticity;
Of the N chromaticities, the total average chromaticity or the multiple average chromaticity
To select and output chromaticity belonging to a limited area around
Output from the area data output section and the limited area data output section.
Calculate the average coordinates of the applied chromaticity as multiple average chromaticity
An area-limited average calculator, the total average chromaticity and the multiple average
A light source chromaticity candidate storage unit for storing and outputting chromaticity.
Chromaticity belonging to a limited area around the multiple average chromaticity
A multi-average chromaticity is repeatedly obtained from the
Item 3. The light source estimation device according to Item 1 . 3. The illuminant chromaticity determination unit determines the presence of the dominant color.
Determination result from parts is the total average chromaticity and priority chromaticity in the case of dominant colors absent, multiple average chromaticity and the priority chromaticity selection unit for a priority chromaticity Otherwise, the multiple average chromaticity is rough The fire
If it is included in the area set on the chromaticity plane,
Constant, and the specific color determination unit determines that the non-specific color when not included in the area, the multiple average chromaticity in priority chromaticity, and, prior to
If the judgment result from the specific color judgment unit is a specific color, its chromaticity
It was a light source chromaticity, the total average chromaticity otherwise
A light source estimating apparatus according to claim 2 comprising a total chromaticity judging unit that the light source chromaticity. 4. The method according to claim 1, wherein the priority chromaticity from the light source chromaticity determination unit is large.
In weighted average chromaticity and when it is determined that the non-Japanese Teiiro, determination
And a determination result from the comprehensive determination unit
If the result can be determined, light source estimation from the light source estimation unit
The light source that holds the result, and if it cannot be determined, it is held inside
A light source estimation result holding unit for outputting the estimation result
Item 3. The light source estimation device according to item 3 . 5. The N chromaticities photograph a color chart.
Whether the color chart image is a distribution on the chromaticity plane
A color chart recognizing unit for recognizing the N chromaticities
For color charts that calculate average coordinates as total average chromaticity
The light source chromaticity estimator and the
If the total chromaticity is the light source chromaticity,
Outputs the light source chromaticity from the light source chromaticity determination unit to the light source estimation unit.
The light source estimation device according to claim 1, further comprising: an estimated light source chromaticity selection unit . 6. The recognition of a color chart is performed by using the N colors.
Shows how much the chromaticity distribution of the degree is along the principal component axis.
The rotation correlation coefficient is small, and the variance of the N chromaticities is
6. The light source estimating device according to claim 5, wherein is larger .