JP5225403B2 - White balance adjustment device and color identification device - Google Patents

Description

  The present invention relates to a technique for adjusting white balance of an image taken by an imaging apparatus such as a digital camera.

  In a digital camera or the like, auto white balance adjustment is performed to reproduce a white subject white. As the white balance method, a method of adjusting the balance of the RGB components of the signal of each pixel so that the average of the entire image becomes an achromatic color is used. Also, the image is divided into a plurality of blocks, the average value of RGB of each block is calculated, only blocks whose average value belongs to a predetermined range are extracted, and the average value of RGB of the extracted block group is There is also known a technique for adjusting each of RGB components so that an achromatic color is obtained.

  Further, the following patent document discloses a technique for dividing an image into a plurality of blocks, calculating a representative value of each block, and calculating a white balance gain using the representative values of all the blocks.

  FIG. 12 shows a configuration block diagram of this prior art. The imaging device 10 such as a digital camera takes an image and outputs the obtained image to the block dividing circuit 12.

  The block dividing circuit 12 equally divides the input image into a plurality of blocks. Each block is composed of n × m pixels. The block division circuit 12 sequentially outputs each block to the representative value calculation circuit 14.

The representative value calculation circuit 14 calculates an RGB average value of n × m pixels constituting each block, and performs the following linear transformation on the average value to obtain representative values (Tl, Tg, Ti). calculate.
Here, Tl is the luminance of the block, and Tg and Ti are the color differences of the block. The representative value calculation circuit 14 outputs the calculated representative values (Tl, Tg, Ti) of each block to the white balance evaluation circuit 16.

  The white balance evaluation circuit 16 evaluates the reliability of each block, calculates a weighting factor corresponding to the reliability, and outputs the weight coefficient to the white balance gain calculation circuit 18. The reliability is evaluated by using, for example, empirical knowledge about each light source. When the luminance of the subject is very high, the reliability is not likely to be a fluorescent lamp. The weight of the block that is highly likely to be) is set smaller as the subject brightness increases.

The white balance gain calculation circuit 18 calculates a white balance gain by performing weighted averaging using a representative value of each block and a weighting factor based on the reliability of each block calculated by the white balance evaluation circuit 16. Specifically, the white balance gain is calculated by the following equation.
Here, TlMix, TgMix, and TiMix are weighted average values of representative values of each block. The (RMix, Gmix, Bmix) calculated by the above formula is the color of the light source that illuminates the subject. The white balance gains Rgain, Ggain, and Bgain are gray when the light source color is reflected by a white object. It is set to correct (that is, R = G = B). The white balance gain calculation circuit 18 outputs the calculated gain to the white balance adjustment circuit 20.

  The white balance adjustment circuit 20 adjusts the white balance of the image by multiplying the pixel values R, G, and B of each pixel constituting the image input from the imaging device 10 by the gain calculated by the white balance gain calculation circuit 18. And output the result.

JP 2000-92509 A

  However, in the above prior art, the R, G, B average values of all the pixels constituting each block are obtained, and the representative value of each block is calculated using equation (1) for the average value. When a specific object color that is not a light source color exists in a block, the average value of the block is affected by the object color.

  FIG. 13 shows a block 100 when the image is divided into a plurality of blocks. The block 100 is composed of n × m pixels, and a simple average of R, G, and B of all the pixels is calculated. However, as shown in the figure, when a green leaf image 102 exists in the block 100, the background color and the green color are averaged, so the average value is the color of the light source illuminating the block. Therefore, there may be a situation in which a place that should be determined as daylight is erroneously determined as a fluorescent lamp.

  As described above, according to the conventional technique, there is not enough countermeasures when a chromatic color is present in a block or when the block is not uniform, which causes a decrease in accuracy of white balance adjustment.

  An object of the present invention is to provide an apparatus capable of reliably performing white balance adjustment even when a chromatic color is present in a block or when the block is not uniform.

The present invention is a white balance adjustment device that adjusts the white balance of an input image, the dividing means for dividing the input image into a plurality of clusters based on the chromaticity of each pixel of the input image, and calculating the representative value of each cluster Representative value calculating means, determining means for determining whether each cluster is an object color cluster based on the representative value, and representative values of remaining clusters excluding clusters determined as object color clusters among all clusters Gain calculating means for calculating a white balance gain using the above-mentioned, and the representative value of the cluster includes a color difference of the cluster, and stores in advance a light source color area and an apparent object color area on the color difference plane And the determination unit is configured to determine whether the color difference included in the representative value of the cluster belongs to an obvious object color area stored in the storage unit. Characterized in that determining said object color cluster.

In addition, the present invention is a color identification device for identifying whether or not a cluster that is composed of a group of pixels having the same chromaticity constituting an input image reflects a light source color, and calculates a color difference between the clusters. A color difference calculating means, a storage means for storing a color difference area of an achromatic object under a plurality of light sources as a light source color area and a color difference area of a specific color object as an object color area on the color difference plane; Processing means for identifying whether or not the cluster reflects a light source color by determining whether a color difference belongs to a light source color area or an object color area stored in the storage means. To do.

  According to the present invention, white balance adjustment accuracy can be improved, and high quality image quality can be obtained.

It is a configuration block diagram of an embodiment. It is a pixel block diagram of an input image. It is explanatory drawing of the light source color area | region on a color difference plane, and a clear object color area | region. It is explanatory drawing which shows a process result. It is a flowchart of an embodiment. It is a block diagram of the configuration of another embodiment. It is explanatory drawing of clustering. It is identification explanatory drawing of a gray zone. It is a flowchart of an embodiment. It is explanatory drawing of a specific color gamut area | region. It is a position adjustment explanatory diagram of a specific color gamut region. It is a block diagram of a conventional apparatus. It is explanatory drawing in which the leaf image exists in a block.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

<First Embodiment>
FIG. 1 is a block diagram showing the configuration of the white balance adjusting apparatus according to this embodiment. The white balance adjustment device of this embodiment is built in a digital camera as an image processing IC having a processor and a memory. The difference from the conventional configuration shown in FIG. 12 is that a color gamut determination circuit 13 is provided in the preceding stage of the representative value calculation circuit 14.

  The block division circuit 12 equally divides the image input from the imaging device 10 into a plurality of blocks, and sequentially outputs each block to the color gamut determination circuit 13.

  The color gamut determination circuit 13 determines whether or not each of the n × m pixels constituting each block is a pixel having an apparent object color based on the color difference, and determines that the pixel has an apparent object color. Identify non-pixels. The color gamut determination circuit 13 identifies each pixel constituting the block for all blocks and outputs the identification result to the representative value calculation circuit 14. The identification result is obtained, for example, by setting a specific flag for a pixel having an obvious object color.

  The representative value calculation circuit 14 basically calculates the representative value of each block in the same manner as the conventional circuit, but the representative value calculation circuit 14 of the present embodiment calculates the average of all the pixels constituting each block as in the conventional case. Instead of calculating the representative value of each block using the value, the representative value of each block is calculated using the average value of only the remaining pixels excluding the pixels identified as the obvious object color. The representative value calculation for each block is performed using equation (1). The difference is that (R, G, B) is not the average of all pixels, but is the average value of pixels that are not identified as clear object colors. The representative value calculation circuit 14 outputs the representative value of each block to the white balance evaluation circuit 16. The subsequent processing is the same as in FIG. 12, and the representative value of each block is weighted and averaged, and the white balance gain (Rgain, Ggain, Bgain) is calculated from the weighted average value using equations (2) to (4). . In the present embodiment, the representative value is calculated by removing the pixel of the obvious object color from each block, and the white balance gain is calculated based on the representative value. Therefore, each block itself is not necessarily an achromatic color. The weighted average of the representative values of each block is processed as an achromatic color. Therefore, an appropriate white balance gain can be calculated even when there is no achromatic color in the input image. Further, since the average value of the remaining pixels in each block is used as the representative value of that block, noise in each block can be absorbed.

  Hereinafter, the processing of the color gamut determination circuit 13 will be described more specifically.

  FIG. 2 shows the pixels constituting each block input from the block dividing circuit 12. Each of the pixel groups 200 constituting the block is composed of R pixels, G pixels, and B pixels. In the present embodiment, the R pixel, the G pixel, and the B pixel are collectively referred to as one pixel 202. The color gamut determination circuit 13 calculates a color difference (Tg, Ti) using each pixel value of the R pixel, the G pixel, and the B pixel constituting each pixel. The color difference can be calculated by the equation (1) using the R pixel value, G pixel value, and B pixel value of each pixel as (R, G, B).

  FIG. 3 shows a color difference plane (Ti-Tg plane). In the color difference plane, a region 204 generally considered as a light source color and a region 206 considered as a clear object color can be uniquely set. A region 204 that is considered to be a light source color is set as a rectangular region in which the sizes of Ti and Tg are less than a predetermined value, and a region that is considered to be an obvious object color is set as a region in which the sizes of Ti and Tg are larger than a predetermined value. The The color gamut determination circuit 13 stores the areas 204 and 206 shown in FIG. 3 in a memory, and to which area the color difference (Tg, Ti) calculated for each pixel constituting each block belongs. Identify If the predetermined values are Tith and Tgth, and the color difference (Tga, Tia) of a certain pixel constituting a certain block is Tga> Tgth and Tia> Tith, the pixel is in an area 206 that is considered to be an obvious object color. Identify as a pixel.

  The achromatic color refers to the color of an object having a substantially constant (flat) reflectance regardless of the wavelength in spectral reflectance, and the light color obtained by reflecting the light source on this object is considered as the light source color. . The achromatic color fluctuates on the color difference plane due to differences in light sources such as daylight, fluorescent light, and tungsten light. The light source color area 204 in FIG. 3 is set as a range in which the achromatic color varies under various light sources. On the other hand, the obvious object color is the color of an object whose spectral reflectance is not flat. If the light source changes, the spectral reflection distribution of this object naturally changes, but even if the light source changes, the color difference does not belong to the light source color region 204 described above. In other words, when the light source changes, the color difference that enters the light source color region 204 is excluded from the “obvious object color” in the present embodiment. However, a gray zone may exist at the boundary between both regions. The gray zone processing will be described later.

  FIG. 4 schematically shows a determination result of a certain block 100 determined by the color gamut determination circuit 13. Among the pixels constituting the block 100, the pixels indicated by the diagonal lines in the drawing are pixels belonging to the region 206 considered to be an obvious object color, that is, the pixels identified by the pixel 208 of the obvious object color. In the representative value calculation of the block 100, the average value of the pixel values of the remaining pixels excluding the hatched pixels is calculated, and the representative value of the block 100 is calculated according to equation (1) using this average value. Since the average value is calculated by removing the pixel 208 of the obvious object color, the influence of the obvious object color is eliminated, and the representative value is also a value reflecting the light source color. Explaining with reference to FIG. 13, the pixel group constituting the green leaf image 102 is excluded, the average value of only the remaining pixel group is calculated, and the representative value of the block 100 is calculated using this average value. It will be.

  FIG. 5 shows a processing flowchart of the present embodiment. First, when an image is input from the imaging device 10 (S101), the image is divided into blocks of a predetermined size (S102). The block size is arbitrary. Each block is sequentially input, and in each block, the color difference (Tg, Ti) is calculated according to the equation (1) using the pixel values (R, G, B) of each pixel constituting the block (S103). The calculated color difference (Tg, Ti) is stored in the memory in association with information for specifying the pixel, for example, a pixel number or pixel coordinate.

  After calculating the color difference (Tg, Ti) of the pixel of interest, it is determined whether or not the color difference (Tg, Ti) belongs to a preset clear object color region 206 (S104). Specifically, as described above, the threshold values Tgth and Tith for identifying the light source color region 206 and the obvious object color region 208 are determined. If it is determined that the color difference (Tg, Ti) of the pixel of interest belongs to the clear object color region 206, a flag is set for the pixel and stored in the memory (S105). On the other hand, if it is determined that it does not belong to the clear object color area 206, the flag is not set. The above process is repeated for all the pixels constituting the block (S106). Then, after executing for all the pixels in a certain block, the pixels other than the pixel for which the flag is set are read from the memory, the average value of the pixel values of the read pixels is calculated, and the calculated average value is (R , G, B), the representative value of the block is calculated according to the equation (1) (S107). The above processing is executed for all the blocks, and the representative values are calculated for all the blocks (S108). After calculating the representative values of all the blocks, the white balance gain is calculated according to (2) to (4) based on the representative value of each block (S109). The calculated white balance gain is multiplied by the pixel value of each pixel constituting the image input from the imaging device 10 to adjust the white balance.

Second Embodiment
FIG. 6 shows a configuration block diagram of the present embodiment. In the first embodiment, the input image is equally divided into blocks of a predetermined size by the block dividing circuit 12, but in this embodiment, clustering is dynamically performed according to the chromaticity of each pixel constituting the image. For this reason, the clustering block division circuit 22, the representative value calculation circuit 24, and the color gamut determination circuit 26 are provided before the white balance evaluation circuit 16.

  The clustering block dividing circuit 22 inputs an image from the imaging device 10 and clusters each pixel based on the chromaticity of each pixel constituting the image. That is, adjacent pixels having the same chromaticity are sequentially grouped (the same label is attached), and the region range of the same chromaticity is defined. Each clustered area is numbered sequentially and stored in the memory. The clustering block division circuit 22 outputs the pixels belonging to each cluster to the representative value calculation circuit 24.

  The representative value calculation circuit 24 calculates the average value of the pixels belonging to each cluster, and calculates the representative value of the cluster according to (1) color using the calculated average value (R, G, B). When the representative value calculation circuit 24 calculates the representative values for all the clusters, the representative value calculation circuit 24 outputs the representative values to the color gamut determination circuit 26.

  Similar to the color gamut determination circuit 13 in FIG. 1, the color gamut determination circuit 26 stores a light source color area 204 and an apparent object color area 206 as shown in FIG. It is determined whether or not the value color difference (Tg, Ti) belongs to the clear object color region 206. The color gamut determination circuit 26 determines whether or not all the clusters have clear object colors, and then outputs only the representative values of the remaining clusters excluding the clear object colors to the white balance evaluation circuit 16. The subsequent processing is the same as in FIG. In the present embodiment, only representative values of clusters belonging to the light source color region 204 contribute to the calculation of the white balance gain.

  FIG. 7 schematically shows processing of the clustering block division circuit 22, the representative value calculation circuit 24, and the color gamut determination circuit 26. The chromaticity of each pixel constituting the input image 1 is calculated and sequentially stored in the memory. Then, the pixels having the same chromaticity are clustered with the same label. In FIG. 7, it is shown in black and white for convenience of explanation, but it should be noted that it is actually red, green, orange, yellow and the like. In the figure, clusters 300, 302, 304, and 306 are illustrated. All the pixels constituting the cluster 300, that is, the pixels having the same label are read from the memory and the average value (R, G, B) is calculated, and the representative value of the cluster 300 is calculated by the equation (1). It is determined whether or not the color difference (Tg, Ti) of the representative values belongs to the clear object color region 206. If not, the light source color region 300 is identified. Similarly, all the pixels constituting the cluster 302 are read from the memory, the average value (R, G, B) is calculated, and the representative value of the cluster 302 is calculated by the equation (1). It is determined whether or not the color difference (Tg, Ti) among the representative values belongs to the clear object color region 206, and if it belongs, it is identified as the clear object color region 302.

  On the other hand, the cluster 304 is a cluster having a chromaticity corresponding to orange, and is located in the boundary region between the obvious object color region 206 and the light source color region 204 and in the gray zone. Judging from the chromaticity alone, it corresponds to orange and can be identified as a light source color region of tungsten. However, the cluster area 304 is surrounded by an obvious object color area 302, and from this, it is determined that the cluster area 304 is also an obvious object color area. Similarly, the cluster 306 is located in the gray zone of the obvious object color region 206 and the light source color region 204, and corresponds to yellow-green when judged from the chromaticity alone, and can be identified as the light source color region of the fluorescent lamp. Therefore, the cluster 306 is also determined to be an obvious object color region.

In this way, when it is difficult to identify whether it is a light source color region or an obvious object color region,
(1) A cluster with close chromaticity exists in the surrounding area. (2) A cluster with close chromaticity can be identified with high accuracy by determining whether it is a light source color region or an obvious object color region. . That is, when it is difficult to identify a certain cluster, that is, when the color difference (Tg, Ti) is near the threshold (Tgth, Tith) (when the difference is within a predetermined error range), If there is a cluster with close chromaticity and the cluster is an obvious object color area, the cluster of interest is identified as an obvious object color area, and if the cluster is not an obvious object color area, Are also identified as light source color regions. When there are no clusters with close chromaticity around them, or when there are multiple clusters with close chromaticity and their identification results are different from each other, it can be arbitrarily set which area to identify. Are identified as distinct object color regions. An example of an algorithm for identifying a focused cluster in the gray zone as an obvious object color area is shown below.

(A) A cluster of interest is identified as an apparent object color region when there is a cluster that is close to chromaticity and already identified as an obvious object color.
(B) If there is no adjacent cluster that has already been identified as an object color that is close to chromaticity, and if the cluster that has already been identified as an object color surrounds the cluster, It is identified as an obvious object color area.

  FIG. 8 schematically shows the above processing in the gray zone. It is assumed that the color difference of the cluster 310 of interest exists in the gray zone on the color difference plane, and the object color region 308 exists around or adjacent to the cluster 310. If the object color region 308 exists adjacent to the cluster 310 that cannot be identified by the cluster 310 alone, the chromaticity of the cluster 310 is compared with the chromaticity of the object color region 308. When the chromaticity difference is within a predetermined allowable range, the cluster 310 is identified as the object color region 310 as shown in the figure. When the chromaticity difference exceeds a predetermined allowable range, it is determined whether or not the object color region surrounds the surroundings, and even if it is surrounded, it is identified as the object color region.

  FIG. 9 shows a processing flowchart of the present embodiment. When an image is input from the imaging device 10 (S201), a chromaticity is calculated for each pixel constituting the input image without being divided into blocks, and a cluster is formed by grouping pixel groups having the same chromaticity. (S202). Clustering is a well-known technique for image processing using a computer. The chromaticity of a certain pixel is calculated and compared with the chromaticity of adjacent pixels. If they match, both pixels are labeled with the same labeling. The above processing is performed for all the pixels, and the pixel group with the same label is numbered as one cluster and stored in the memory. Pixel 1, 2, 3 belongs to cluster 1, pixels 4, 5, 6, 7 belong to cluster 2, and so on. After clustering, the representative value of the cluster is calculated using the pixel values of the pixels belonging to each cluster (S203). That is, the average value of the pixels belonging to the cluster is calculated, and the luminance Tl and the color difference (Tg, Ti) of the cluster are calculated by the equation (1) using the average value.

  After calculating the representative values (Tl, Tg, Ti) of the clusters, the color difference (Tg, Ti) of the representative values is set in advance on the color difference plane, and the light source color region 204 and the obvious object color region 206 (see FIG. 3). And whether it belongs to the clear object color area 206 or not (S204). If it belongs to the obvious object color area 206, a flag is set for the cluster (S205). If it does not belong to the obvious object color area 206, no flag is set for that cluster.

  If the color difference (Tg, Ti) of the cluster is located in the vicinity of the boundary (gray zone) between the light source color region 204 and the obvious object color region 206 in the determination processing of S205, the chromaticity is next adjacent to the cluster. It is determined whether or not there is a cluster identified as an apparent object color area. If it exists, the flag is set on the assumption that the cluster also belongs to the clear object color region. If it does not exist, it is further determined whether or not the surrounding area is surrounded by a clear object color area, and if it is surrounded, a flag is set as belonging to the clear object color area. In FIG. 3, a gray zone having a predetermined width may be provided at the boundary between the light source color region 204 and the obvious object color region 206. When the color difference of the cluster is located in this gray zone, the process proceeds to the above determination process.

  The above processing is repeatedly executed for all clusters (S206), and after identifying all the clusters, a white balance gain is calculated using the representative values of the remaining clusters excluding the cluster with the flag set (S207).

  Also according to the present embodiment, the white balance gain is calculated by excluding clusters recognized as clear object colors, so that the white balance adjustment accuracy can be improved.

  In the present embodiment, the identification of the cluster located in the gray zone has been described. However, the same processing can be performed on each pixel by the color gamut determination circuit 13 in the first embodiment. When determining whether each pixel is the light source color area 204 or the obvious object color area 206 based on the color difference of each pixel, pay attention when surrounding the area with clear object color pixels. The identified pixel is also identified as an object color pixel.

<Third Embodiment>
In the above embodiment, it is determined whether or not each pixel or each cluster belongs to the clear object color area 206 using the area map shown in FIG. 3, but a specific object-specific color difference area is set. It is also possible to improve the identification accuracy by determining whether or not it belongs to these specific areas.

  FIG. 10 shows an example of the specific color gamut region. As the specific region, the blue sky region 400, the skin color region 402, the leaf region 404, and the flower region 406 are mapped on the color difference plane. When the color difference of the cluster belongs to these specific color gamut regions, it is identified as an obvious object color region. For example, when the color difference of a certain cluster belongs to the leaf area 404 or the flower area 406, the cluster is identified as an obvious object color area. Even if a certain cluster exists in the gray zone, if it belongs to these specific color gamut regions, the cluster can be identified as an apparent object color region and excluded from the calculation of the white balance gain.

<Fourth embodiment>
In the third embodiment, the identification accuracy is improved by mapping specific color gamut regions on the color difference plane and determining whether or not the color difference of the cluster belongs to these specific color gamut regions. Depends on the mapping accuracy of each specific color gamut region. The specific color gamut is a skin color area 402 or a leaf area 404 as shown in FIG. 10, and the identification accuracy improves as the specific color gamut area increases, while more accurate mapping is performed as the specific color gamut area is subdivided. It becomes difficult.

  Therefore, in this embodiment, an accurate mapping position of another specific color gamut region is estimated from the positional relationship of the subject in a specific color gamut region.

  FIG. 11 schematically shows the processing of this embodiment. The subject used for mapping correction is a human face image. There are eyes in the face image and lips. Here, it is assumed that there are two face images 500 and 502 with different white balance. The color difference between the skin color portion and lip portion of the face image 500 and the skin color portion and lip portion of the face image 502 are different. On the other hand, on the color difference plane, a skin color region 504 is mapped in consideration of daylight (daylight), a fluorescent lamp, and a tungsten light source. The color difference of the skin color portion in the face image 500 is located anywhere in this skin color area 504 (position P), and the color difference of the skin color image in the face image 502 is also located anywhere in the skin color 504 (position Q). . However, both positions are at different positions in the skin color region 504 because the white balance of the original image is different. Since the lip part exists in the skin color part, the position of the lip part naturally differs if the position of the skin color part is different. Therefore, as shown in the figure, when the lip color area is mapped as the specific color gamut area, the lip color area 506 is set according to the position P in the skin color area 504, or according to the position Q in the skin color area 504. The skin color area 508 is set. That is, the position of the lip color region is shifted according to the color difference position of the skin color portion of the image in the skin color region 504. Specifically, the processor of the image processing IC stores the lips stored in the same memory in accordance with the deviation amount between the center position of the skin color area 504 stored in the memory and the position P (or position Q) of the skin portion. Shift the default position of the color area and store it again in memory. As a result, the lip color region can be accurately mapped, and the identification accuracy of whether or not a certain cluster is a lip color is improved. In the figure, the lip portion of the face image 500 is located in the lip color area 506 set on the color difference plane according to the position P, and the lip portion of the face image 502 is set on the color difference plane according to the position Q. The lip color region 508 is positioned. It will be understood that when only the lip color area 506 exists, the lip portion of the face image 502 is outside the lip color area 506 and cannot be accurately identified. The setting of the specific color gamut region according to the positions P and Q can be similarly applied not only to the lip color region but also to other specific color gamut regions. For example, the eye part is composed of a black eye part and a white eye part, but the position of the specific color region of the achromatic white eye part can be variably set according to the positions P and Q. In the case of the eye part, it is possible to further improve the accuracy by paying attention not only to the color difference but also to the luminance difference between the white eye part and the black eye part.

  10 imaging device, 12 block division circuit, 13 color gamut determination circuit, 14 representative value calculation circuit, 16 white balance evaluation circuit, 18 white balance gain calculation circuit, 20 white balance adjustment circuit.

Claims (9)

A white balance adjustment device for adjusting the white balance of an input image,
Dividing means for dividing the input image into a plurality of clusters based on the chromaticity of each pixel of the input image;
Representative value calculating means for calculating the representative value of each cluster;
Determining means for determining whether each cluster is an object color cluster based on the representative value;
A gain calculating means for calculating a white balance gain using a representative value of remaining clusters excluding a cluster determined as an object color cluster among all the clusters;
Have,
The representative value of the cluster includes the color difference of the cluster,
Storage means for preliminarily storing a light source color area and a clear object color area on a color difference plane
And the determination unit determines that the color difference included in the representative value of the cluster belongs to the clear object color region stored in the storage unit, and determines the object color cluster. Balance adjustment device. The apparatus of claim 1 .
The determination unit, when the color difference to be determined is in the vicinity of the border of the apparent object color area and the light source color region, the white balance adjustment and judging according to the determination result of the surrounding cluster apparatus. The apparatus of claim 1 .
The storage means further stores in advance a specific color region specific to the specific object,
The determination means, the color difference to be determined is the white balance adjustment apparatus characterized by determining the object Ilok raster when belonging to a specific color region which is stored in the storage means. The apparatus of claim 3 .
The positions of the plurality of specific color areas on the color difference plane are set in accordance with a deviation between the position of the specific part color of the input image and the center position of the reference specific color area in a certain reference specific color area. White balance adjustment device characterized by. The apparatus of claim 4 .
2. The white balance adjusting apparatus according to claim 1, wherein the reference specific color region is a skin color region under various light sources, and the specific part color of the input image is a lip color. A color identification device that identifies whether or not a cluster composed of pixel groups having the same chromaticity constituting an input image reflects a light source color,
Color difference calculating means for calculating the color difference of the cluster;
Storage means for storing a color difference area of an achromatic object under a plurality of light sources as a light source color area on the color difference plane and storing a color difference area of a specific color object as an object color area;
Processing means for identifying whether the cluster reflects the light source color by determining whether the color difference of the cluster belongs to a light source color area or an object color area stored in the storage means;
A color identification device characterized by comprising: The apparatus of claim 6 .
The color identification apparatus, wherein the object color area includes at least one of a skin color area, a green area of leaves, a red area of flowers, and a blue area of blue sky. The apparatus of claim 6 .
The object color area includes a skin color area,
A color identification apparatus comprising: means for variably setting at least one position of the object color area in accordance with a position of a color difference of a skin color cluster of the input image in the skin color area. The apparatus of claim 6 .
The processing means is further configured such that when the color difference of the cluster to be determined is near the boundary between the light source color area and the object color area , the chromaticity is already identified as an object color area adjacent to the determination target cluster. A color identification device for identifying whether or not the cluster reflects a light source color based on whether or not the cluster exists.