JP2013168723A - Image processing device, imaging device, image processing program, and image processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technique capable of quickly and accurately discriminating a scene of a subject without increasing a calculation amount or a circuit scale.
SOLUTION: Using a feature amount of a target image, a plurality of determination means for respectively obtaining a similarity between a preset color gamut of each reference scene and a shooting scene of the target image, and obtained from the plurality of determination means Specifying means for specifying a coordinate point on a color space corresponding to a shooting scene in accordance with a correlation between a plurality of similarities.
[Selection] Figure 1

Description

  The present invention relates to an image processing device, an imaging device, an image processing program, and an image processing method.

  Conventionally, various techniques for identifying a subject scene have been developed in order to set parameter values in image processing such as exposure conditions and white balance processing according to the subject scene.

  For example, by inputting shooting information such as a feature amount such as a color average of image data and an aperture value to a support vector machine for each scene that determines whether or not it belongs to a predetermined scene, the scene of the subject in the image data is determined. There is a technique for identifying (see Patent Document 1).

JP 2008-228087 A

  However, in the prior art, it is necessary to prepare support vector machines and data for discrimination criteria for each scene to be identified, which increases the amount of calculation and the circuit scale.

  SUMMARY OF THE INVENTION In view of the above-described problems of the prior art, an object of the present invention is to provide a technique capable of quickly and accurately discriminating a subject scene without increasing the amount of calculation and the circuit scale.

  In order to solve the above-described problem, an aspect of the image processing apparatus illustrating the present invention uses a feature amount of the target image to resemble a preset color gamut of each reference scene and a shooting scene of the target image. A plurality of discriminating means for obtaining the degrees, and a specifying means for identifying the coordinate point on the color space corresponding to the shooting scene according to the correlation of the plurality of similarities acquired from the plurality of discriminating means.

  Further, the specifying unit may specify a coordinate point on the color space corresponding to the photographic scene according to the feature amount of the target image and the correlation between the plurality of similarities.

  The specifying unit may determine a distribution according to the correlation of a plurality of input values on the color space, and specify the coordinate point based on the center of gravity of the calculated distribution or the position giving the maximum value in the distribution.

  The specifying means may be a neural network.

  Further, the reference scene may be a predetermined light source, and the shooting scene may be an illumination light source in the target image.

  Further, an adjustment unit that obtains an adjustment amount of white balance for the target image based on the specified coordinate point may be provided.

  One aspect of an imaging apparatus illustrating the present invention includes an imaging unit that captures an image of a subject and generates a target image, and the image processing apparatus of the present invention.

  An aspect of the image processing program exemplifying the present invention is a determination procedure for determining a similarity between a preset color gamut of each reference scene and a shooting scene of the target image, using the feature amount of the target image. In accordance with the correlation of the plurality of similarities acquired in the procedure, the computer is caused to execute as a specific procedure for specifying coordinate points on the color space corresponding to the shooting scene.

  An aspect of the image processing method illustrating the present invention includes a determination step for obtaining a similarity between a preset color gamut of each reference scene and a shooting scene of the target image using the feature amount of the target image, A specifying step of specifying coordinate points on the color space corresponding to the shooting scene according to the correlation of the plurality of similarities acquired in the determination step.

  According to the present invention, it is possible to quickly and accurately discriminate a scene of a subject without increasing a calculation amount or a circuit scale.

The block diagram which shows the structure of the digital camera of one Embodiment The figure which shows the structure of the discrimination | determination part 21 Diagram showing illumination color distribution of each light source The figure which shows the structure of the specific | specification part 22 6 is a flowchart illustrating an image processing operation performed by the digital camera according to the embodiment. The flowchart which shows the operation | movement of the image processing by the digital camera of other embodiment.

<< One Embodiment >>
FIG. 1 is a block diagram illustrating an example of a configuration of a digital camera 1 according to an embodiment.

  The digital camera 1 according to the present embodiment includes an imaging optical system 11, an imaging element 12, a DFE 13, a CPU 14, a memory 15, an operation unit 16, a monitor 17, and a media interface (I / F) 18. The DFE 13, the memory 15, the operation unit 16, the monitor 17, and the media I / F 18 are each connected to the CPU 14.

  The image sensor 12 is a device that captures an image of a subject formed by a light beam that has passed through the imaging optical system 11. The output of the image sensor 12 is connected to the DFE 13. Note that the image sensor 12 of the present embodiment may be a progressive scanning solid-state image sensor (CCD or the like) or an XY address type solid-state image sensor (CMOS or the like).

  A plurality of light receiving elements are arranged in a matrix on the light receiving surface of the imaging element 12. In each light receiving element of the image sensor 12, red (R), green (G), and blue (B) color filters are arranged according to a known Bayer array. Therefore, each light receiving element of the imaging element 12 outputs an image signal of a RAW image corresponding to each color by color separation with a color filter. Thereby, the image sensor 12 can acquire a color RAW image at the time of imaging.

  In addition, the image sensor 12 captures a still image or a main image of a moving image in response to a full press operation of a release button included in the operation unit 16 in the shooting mode of the digital camera 1. Further, the imaging element 12 in the shooting mode captures composition confirmation images (through images) at predetermined intervals even during standby for shooting. The through image data is output from the image sensor 12 by thinning-out readout.

  The DFE 13 is a digital front-end circuit that performs signal processing such as A / D conversion of image signals input from the image sensor 12 and correction of defective pixels. In this embodiment, the DFE 13 constitutes an image pickup unit together with the image pickup device 12 and outputs an image signal input from the image pickup device 12 to the CPU 14 as image data.

  The CPU 14 is a processor that comprehensively controls each unit of the digital camera 1. For example, the CPU 14 performs autofocus (AF) control by known contrast detection, known automatic exposure (AE) calculation, and the like based on the output of the image sensor 12. Further, the CPU 14 performs digital processing on the image data of the RAW image from the DEF 13. Digital processing includes, for example, interpolation processing, white balance processing, gradation conversion processing, contour enhancement processing, and color conversion processing.

  Further, the CPU 14 according to the present embodiment operates as a calculation unit 20, a determination unit 21, a specification unit 22, and an adjustment unit 23 by executing the image processing program. In the present embodiment, the processing of the calculation unit 20, the determination unit 21, the specification unit 22, and the adjustment unit 23 is performed on the RAW image before the digital processing is performed. The CPU 14 performs digital processing, particularly white balance processing, on the RAW image based on the processing result.

  The computing unit 20 uses, for example, image feature values such as luminance (BV), R / G, and B / G average values and standard deviation values from a RAW image (target image) captured by the image sensor 12 and output from the DFE 13. Is calculated. In the present embodiment, the image feature quantity and the shooting feature quantity such as the focal length of the imaging optical system 11 at the time of shooting the target image are collectively referred to as a “feature quantity vector”.

  As shown in FIG. 2, the determination unit 21 includes, for example, four determination units of known support vector machines (SVM) 24-1 to 24-4. Each SVM 24 has color gamut data of a reference scene calculated in advance by supervised learning for discriminating a subject scene (captured scene) of the target image, and the scene of the target image using the feature quantity vector. The degree of similarity (similarity) between the reference scene and the reference scene is calculated.

  Note that the determination unit 21 of the present embodiment determines an illumination light source in the target image, which is important in the white balance process, as a scene of the target image. Then, as the reference scene of the present embodiment, for example, there are four light sources: clear sky, shade, light bulb, and fluorescent light. Therefore, each SVM 24 of the determination unit 21 has data on the color gamut of the illumination color of each light source, and calculates the similarity between the illumination light source in the target image and each of clear sky, shade, light bulb, and fluorescent lamp.

  The specifying unit 22 specifies the illumination light source in the target image according to the similarity between each reference scene and the correlation between a plurality of input values of the feature amount vectors by the determining unit 21. For this purpose, the specifying unit 22 of the present embodiment performs, for example, a known neural network.

  Here, using the distribution of the illumination color of each light source shown in FIG. 3 and the data flow diagram in the neural network of the specifying unit 22 shown in FIG. 4, the relationship between the processing of the determination unit 21 and the specifying unit 22 in this embodiment. Will be described.

  FIG. 3 shows an example of the illumination color distribution of each light source such as clear sky, shade, light bulb, and fluorescent light defined in this embodiment on the coordinate plane of the color difference CrCb. As shown in FIG. 3, in general, the distribution of the light sources is not an independent relationship but is distributed in an overlapping manner. That is, for example, fluorescent lamps include fluorescent lamps that emit white light or light bulb color, and fluorescent lamps that emit reddish or bluish light. In the shade, there are various situations such as cloudy weather and indoors in fine weather. This seriously affects white balance processing and the like, and also affects the appearance of the image.

  However, in the prior art, for example, in order to discriminate between fluorescent light emitting white light, light bulb color, reddish or bluish light, it is necessary to prepare dedicated SVM and color gamut data. As a result, the calculation amount and the storage capacity increase and the cost increases. On the other hand, when a neural network is used, the above-described fluorescent lamp can be discriminated with a single neural network, but it is inferior in generalization performance as compared with SVM.

  Therefore, in the present embodiment, first, the SVMs 24-1 to 24-4 of the determination unit 21 calculate the similarity between the illumination light source and the clear sky, shade, light bulb, and fluorescent lamp in the target image. Then, the specifying unit 22 inputs the similarity and the feature amount vector to the neural network, determines whether the fluorescent light is a white light or a light bulb-colored fluorescent light, or whether it is cloudy or indoors in the shade, and the target image. The illumination light source at is specified. As described above, in this embodiment, by combining the SVMs 24-1 to 24-4 of the determination unit 21 and the neural network of the specifying unit 22, it is possible to perform the same processing as when SVMs for various light sources are prepared. become.

  Here, the perceptron of the output layer of the neural network shown in FIG. 4 corresponds to each of the areas obtained by dividing the coordinate plane of the color difference CrCb shown in FIG. 3 by a predetermined size. However, FIG. 4 shows only a partial region of 5 × 5. Each perceptron in the intermediate layer can output a signal to all regions of the perceptron in the output layer, and FIG. 4 shows a part of the output. Note that the size of the region is preferably determined according to the specific accuracy of the required illumination light source.

  In addition, the connection state and the connection coefficient of the input layer, the intermediate layer, and the output layer of the neural network illustrated in FIG. 4 are used as the similarity and the feature amount vector by each SVM 24 in a plurality of images captured by a plurality of known illumination light sources. Suppose that it is determined in advance based on the correlation between the known illumination light source and the clear sky, shade, light bulb, and fluorescent lamp.

  The specifying unit 22 calculates, for example, a weighted average of the distribution of illumination colors output from the neural network on the coordinate plane of the color difference CrCb shown in FIG. The specifying unit 22 specifies the illumination light source in the target image from the coordinate position of the center of gravity.

  The adjustment unit 23 obtains the gain value of each color component of RGB by a known method from the value of the color difference CrCb at the coordinate position of the center of gravity obtained by the specifying unit 22, and calculates the correction value in the white balance process.

  The memory 15 is a non-volatile semiconductor memory that stores image data such as a RAW image, color gamut data of the SVMs 24-1 to 24-4, an image processing program executed by the CPU 14, and the like.

  For example, the operation unit 16 receives an input of an imaging mode switching setting, an instruction to capture a still image or a moving image, and the like from the user.

  The monitor 17 is a monitor such as a liquid crystal monitor and displays various images according to control instructions from the CPU 14. For example, the monitor 17 displays an image in accordance with a control instruction from the CPU 14 after imaging or image processing.

  A non-volatile storage medium 19 can be detachably connected to the media I / F 18. The media I / F 18 executes data writing / reading with respect to the storage medium 19. The storage medium 19 includes a hard disk, a memory card incorporating a semiconductor memory, or the like. In FIG. 1, a memory card is illustrated as an example of the storage medium 19.

  Next, the processing operation by the digital camera 1 of the present embodiment will be described with reference to the flowchart of FIG. In the following description, the target image is a RAW image of a still image, but the same processing is applied to a frame of a moving image or a through image.

  When the CPU 14 receives, for example, an instruction to capture a still image (for example, full pressing operation of a release button included in the operation unit 16) from the user, the CPU 14 causes the image sensor 12 to capture a still image. CPU14 starts the process from step S101 with respect to the captured still image.

  Step S101: The CPU 14 reads a RAW image of a still image output from the image sensor 12 via the DEF 13. The CPU 14 acquires and holds the RAW image exposure conditions, the focal length of the image pickup optical system 11 at the time of image pickup, and the like as shooting feature amounts.

  Step S102: The calculation unit 20 of the CPU 14 uses the read RAW image as a target image, and calculates image feature amounts such as luminance (BV), R / G, B / G average values and standard deviation values.

  Step S103: The determination unit 21 of the CPU 14 inputs the feature amount vector to each of the SVMs 24-1 to 24-4, and calculates the similarity between the illumination light source and each of the light sources of the clear sky, shade, light bulb, and fluorescent lamp in the target image. To do.

  Step S104: The specifying unit 22 of the CPU 14 inputs the similarity and the feature amount vector calculated in Step S103 to the neural network, and obtains the illumination color distribution on the coordinate plane of the color difference CrCb. The specifying unit 22 obtains the position of the center of gravity of the distribution and specifies the illumination light source of the target image.

  Step S105: The adjusting unit 23 of the CPU 14 obtains the gain value of each color component of RGB by a known method from the value of the color difference CrCb at the coordinate position of the center of gravity obtained in step S104, and calculates the correction value in the white balance process. To do.

  Step S106: The CPU 14 performs digital processing such as white balance processing on the target image using the correction amount calculated in step S105.

  Step S107: The CPU 14 records the digitally processed still image in the memory 15 and the storage medium 19 together with the RAW image. Further, the CPU 14 displays a still image on which the digital processing has been performed on the monitor 17. The CPU 14 ends a series of processes.

  As described above, in the present embodiment, by using the SVMs 24-1 to 24-4 of the determination unit 21 and the neural network of the specifying unit 22 together, the illumination light source in the target image can be specified with high accuracy.

Further, by using a small number of SVMs 24-1 to 24-4, it is possible to specify the illumination light source of the target image at high speed with a small amount of calculation. Furthermore, since there is little color gamut data, an increase in circuit scale can be avoided and cost reduction can be achieved.
<< Modification of one embodiment >>
A digital camera according to a modification of the embodiment of the present invention is the same as the digital camera 1 according to the embodiment shown in FIG. Therefore, a detailed description of the configuration of the digital camera 1 of the present embodiment is omitted.

  Further, the processing operation by the digital camera 1 of the present embodiment is the same as the processing of the first embodiment shown in FIG.

  The difference between the digital camera 1 according to the present embodiment and that according to the first embodiment is that the specifying unit 22 performs the processing by the neural network a plurality of times. That is, the specifying unit 22 of the present embodiment feeds back the output of the neural network as an input of its own neural network, and performs the process again together with the feature amount vector. The specifying unit 22 specifies the illumination light source in the target image by performing this feedback processing a plurality of times. Each time the feedback is performed, the neural network of the specifying unit 22 changes, for example, the combination and coupling coefficient between the perceptrons of the input layer, the intermediate layer, and the output layer, increases or decreases the number of perceptrons of the intermediate layer, It is preferable to change the size of the region that is the perceptron of the output layer.

  As described above, in the present embodiment, by using the SVMs 24-1 to 24-4 of the determination unit 21 and the neural network of the specifying unit 22 together, the illumination light source in the target image can be specified with high accuracy.

  Further, the illumination light source can be specified with higher accuracy by feeding back the output from the neural network.

Further, by using a small number of SVMs 24-1 to 24-4, it is possible to specify the illumination light source of the target image at high speed with a small amount of calculation. Furthermore, since there is little color gamut data, an increase in circuit scale can be avoided and cost reduction can be achieved.
<< Other embodiments >>
A digital camera according to another embodiment of the present invention is the same as the digital camera 1 according to one embodiment shown in FIG. Therefore, a detailed description of the configuration of the digital camera 1 of the present embodiment is omitted.

  The difference between the digital camera 1 according to the present embodiment and that according to the first embodiment is that the illumination light source in the through image captured before capturing the main image is specified, and the result is the illumination light source in the main image. It is in the point used for specific.

  Next, the processing operation by the digital camera 1 of the present embodiment will be described with reference to the flowchart of FIG. In the following description, the main image is assumed to be a still image, but the same processing is applied to a moving image frame.

  For example, when the user presses a power button included in the operation unit 16, the CPU 14 turns on and initializes the digital camera 1. For example, when the shooting mode is set, the CPU 14 causes the imaging device 12 to capture a through image of the subject at a predetermined interval during shooting standby until receiving an imaging instruction for the main image from the user. CPU14 starts the process from step S201 with respect to the captured through image.

  Step S201: The CPU 14 reads through images that are read out and output from the image sensor 12 via the DEF 13. The CPU 14 acquires and holds the exposure condition of the through image, the focal length of the imaging optical system 11 at the time of imaging, and the like as the imaging feature amount.

  Step S202: The calculation unit 20 uses the read through image as a target image, and calculates image feature amounts such as luminance (BV), R / G, B / G average value and standard deviation value.

  Step S203: The determination unit 21 inputs the feature amount vector to each of the SVMs 24-1 to 24-4, and calculates the similarity between the illumination light source in the through image and each of the light sources of clear sky, shade, light bulb, and fluorescent lamp.

  Step S204: The specifying unit 22 inputs the similarity and the feature amount vector calculated in Step S203 to the neural network, and obtains the illumination color distribution on the coordinate plane of the color difference CrCb. The specifying unit 22 obtains the position of the center of gravity of the distribution and specifies the illumination light source of the through image.

  Step S205: The adjusting unit 23 obtains the gain value of each color component of RGB by a known method from the value of the color difference CrCb at the coordinate position of the center of gravity of the through image obtained in step S204, and obtains the correction value in the white balance processing. calculate.

  Step S206: The CPU 14 performs digital processing such as white balance processing on the through image using the correction amount calculated in step S205.

  Step S207: The CPU 14 displays the digitally processed through image on the monitor 17, and records the illumination color distribution for the illumination light source of the through image obtained in step S204 in an internal memory (not shown).

  Step S208: The CPU 14 determines whether or not an imaging instruction for the main image has been received, for example, by a full pressing operation of a release button included in the operation unit 16 of the user. When the CPU 14 receives an imaging instruction, the CPU 14 proceeds to step S209 (YES side). On the other hand, when the imaging instruction has not been received, the CPU 14 proceeds to step S201 (NO side), and performs the processing of step S201 to step S207 on the next captured through image.

  Step S209: The CPU 14 reads the RAW image of the still image of the main image output from the image sensor 12 via the DEF 13. The CPU 14 acquires and holds shooting feature amounts such as the exposure condition of the RAW image and the focal length of the imaging optical system 11 at the time of imaging.

  Step S210: The calculation unit 20 uses the read RAW image as a target image, and calculates image feature amounts such as luminance (BV), R / G, B / G average values and standard deviation values.

  Step S211: The determination unit 21 inputs the feature vector into each of the SVMs 24-1 to 24-4, and calculates the similarity between the illumination light source in the RAW image and each of the light sources of the clear sky, shade, light bulb, and fluorescent lamp.

  Step S212: The identifying unit 22 inputs the illumination color distribution data for the illumination light source of the latest through image calculated in step S204, together with the similarity and the feature vector calculated in step S211 to the neural network. The specifying unit 22 determines the illumination color distribution for the illumination light source of the RAW image and the position of the center of gravity of the distribution, and specifies the illumination light source of the RAW image.

  Step S213: The adjustment unit 23 obtains a gain value of each color component of RGB by a known method from the value of the color difference CrCb at the coordinate position of the center of gravity obtained in step S212, and calculates a correction value in the white balance process.

  Step S214: The CPU 14 performs digital processing such as white balance processing on the RAW image using the correction amount calculated in step S213.

  Step S215: The CPU 14 records the still image subjected to the digital processing in the memory 15 or the storage medium 19. Further, the CPU 14 displays a still image on which the digital processing has been performed on the monitor 17. The CPU 14 ends a series of processes.

  As described above, in the present embodiment, by using the SVMs 24-1 to 24-4 of the determination unit 21 and the neural network of the specifying unit 22 together, the illumination light source in the target image can be specified with high accuracy.

  Further, by using the illumination color distribution for the illumination light source of the latest through image, the illumination light source in the main image can be specified more accurately.

Further, by using a small number of SVMs 24-1 to 24-4, it is possible to specify the illumination light source of the target image at high speed with a small amount of calculation. Furthermore, since there is little color gamut data, an increase in circuit scale can be avoided and cost reduction can be achieved.
<< Modification of other embodiment >>
A digital camera according to a modification of the other embodiment of the present invention is the same as the digital camera 1 according to the embodiment shown in FIG. Therefore, a detailed description of the configuration of the digital camera 1 of the present embodiment is omitted.

  The processing operation by the digital camera 1 of the present embodiment is the same as the processing of the other embodiments shown in FIG.

  The difference between the digital camera 1 according to this embodiment and that according to the other embodiment is that the specifying unit 22 performs processing by the neural network a plurality of times, as in the case of the modification of one embodiment. is there. However, it is preferable that the feedback process is not performed for the through image or is less than the number of times for the main image.

  As described above, in the present embodiment, by using the SVMs 24-1 to 24-4 of the determination unit 21 and the neural network of the specifying unit 22 together, the illumination light source in the target image can be specified with high accuracy.

  Further, by using the illumination color distribution for the illumination light source of the latest through image, the illumination light source in the main image can be specified more accurately.

  Furthermore, it is possible to specify with higher accuracy by feeding back the output from the neural network.

Further, by using a small number of SVMs 24-1 to 24-4, it is possible to specify the illumination light source of the target image at high speed with a small amount of calculation. Furthermore, since there is little color gamut data, an increase in circuit scale can be avoided and cost reduction can be achieved.
<< Additional items of embodiment >>
(1) In the above-described embodiment, an example in which the CPU 14 implements the processing of the calculation unit 20, the determination unit 21, the specification unit 22, and the correction unit 23 of the digital camera 1 by software has been described. Of course, these processes may be realized by hardware.

  (2) The image processing apparatus of the present invention is not limited to the example of the digital camera 1 of the above embodiment. For example, a computer in which an image processing program stored in a storage medium or the like is installed in advance causes a captured image to be read directly from the digital camera 1 or via the Internet, and the scene determination process is executed, thereby causing the computer to May function as the image processing apparatus of the present invention.

  (3) In the above embodiment, the calculation unit 20 calculates the image feature amount of the entire target image. However, the present invention is not limited to this, and the image feature amount of only a predetermined partial image region may be calculated. The image feature amount may be calculated for each of a plurality of image regions.

  (4) In the above-described embodiment, the determination unit 21 includes the SVM 24-1 to 24-4 of clear sky, shade, light bulb, and fluorescent lamp, but the present invention is not limited to this, and the SVM of one or more light sources. The number of SVMs is preferably determined as appropriate according to the required specific accuracy, processing speed, storage capacity, and the like.

  The discrimination target of the discrimination unit 21 is not limited to a light source, and may be a scene such as a night view or a portrait.

  (5) In the above embodiment, the specification unit 22 specifies the illumination light source in the target image based on the coordinate position of the center of gravity of the illumination color distribution by the neural network, but the present invention is not limited to this. For example, the specifying unit 22 may specify the illumination light source of the target image based on the coordinate position that gives the maximum value among the illumination color distributions. Alternatively, the specifying unit 22 uses a coefficient based on the determination reliability of each of the SVMs 24-1 to 24-4 of the determination unit 21 to perform weighted average of the illumination color distribution and specify the illumination light source of the target image. Also good. Note that the reliability of the SVM discrimination is the accuracy rate of the SVM discrimination result using a plurality of sample images in which the illumination light source is specified in advance.

  (6) In the above embodiment, the illumination light source in the target image is specified based on the SVMs 24-1 to 24-4 of the determination unit 21 and the neural network of the specification unit 22, but the present invention is not limited to this. For example, the determination unit 21 may use another supervised learning determination method such as boosting instead of SVM. Further, the specifying unit 22 may use template matching or the like instead of the neural network.

  (7) In the above embodiment, one neural network is arranged in the specifying unit 22, but the present invention is not limited to this. For example, the specifying unit 22 may include, for example, a plurality of neural networks having different combinations and coupling coefficients between the perceptrons in the input layer, the intermediate layer, and the output layer, or the number of perceptrons in the intermediate layer. Then, the specification unit 22 calculates the illumination color distribution by inputting the feature vector of the target image, the similarity of the determination unit 21, and the illumination color distribution data with respect to the illumination light source of the through image to each neural network. . The specifying unit 22 may specify the illumination light source of the target image based on the distribution of the illumination colors.

  The specifying unit 22 may have a configuration in which a plurality of other methods such as template matching are used in combination with the neural network.

  From the above detailed description, features and advantages of the embodiments will become apparent. It is intended that the scope of the claims extend to the features and advantages of the embodiments as described above without departing from the spirit and scope of the right. Further, any person having ordinary knowledge in the technical field should be able to easily come up with any improvements and changes, and there is no intention to limit the scope of the embodiments having the invention to those described above. It is also possible to use appropriate improvements and equivalents within the scope disclosed in.

DESCRIPTION OF SYMBOLS 1 ... Digital camera, 11 ... Imaging optical system, 12 ... Imaging device, 13 ... DEF, 14 ... CPU, 15 ... Memory, 16 ... Operation part, 17 ... Monitor, 18 ... Media I / F, 19 ... Storage medium, 20 ... Calculation unit, 21 ... Distinction unit, 22 ... Specification unit, 23 ... Correction unit, 24-1 to 24-4 ... SVM

Claims (9)

A plurality of discriminating means for determining the similarity between the color gamut of each preset reference scene and the shooting scene of the target image using the feature amount of the target image;
A specifying unit for specifying a coordinate point on a color space corresponding to the shooting scene according to the correlation of the plurality of similarities acquired from the plurality of determining units;
An image processing apparatus comprising: The image processing apparatus according to claim 1.
The image processing apparatus characterized in that the specifying unit specifies a coordinate point on a color space corresponding to the shooting scene, based on a feature amount of the target image and a correlation between the plurality of similarities. The image processing apparatus according to claim 1 or 2,
The specifying unit determines a distribution according to the correlation of a plurality of input values on the color space, and specifies the coordinate point based on a position of the center of gravity of the calculated distribution or a position giving a maximum value in the distribution. An image processing apparatus. The image processing apparatus according to any one of claims 1 to 3,
The image processing apparatus according to claim 1, wherein the specifying means is a neural network. The image processing apparatus according to any one of claims 1 to 4,
The reference scene is a predetermined light source,
The imaging scene is an illumination light source in the target image. The image processing apparatus according to any one of claims 1 to 5,
An image processing apparatus comprising: an adjustment unit that obtains an adjustment amount of white balance for the target image based on the identified coordinate point. An imaging unit that images a subject and generates a target image;
An image processing apparatus according to claim 1;
An imaging apparatus comprising: A determination procedure for determining the similarity between the color gamut of each preset reference scene and the shooting scene of the target image using the feature amount of the target image,
A specifying procedure for specifying a coordinate point on a color space corresponding to the shooting scene according to the correlation of the plurality of similarities acquired in the determination procedure;
An image processing program that is executed by a computer. Using the feature amount of the target image, a determination step for obtaining a similarity between the color gamut of each preset reference scene and the shooting scene of the target image,
A specifying step of specifying coordinate points on a color space corresponding to the shooting scene according to the correlation of the plurality of similarities acquired in the determining step;
An image processing method comprising:

Images