JP2009273117A - Image processing system, image processing method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing system capable of reducing a calculation amount when an image is compressed by using a model. <P>SOLUTION: This image processing system includes: a feature area detection part which detects a feature area from a photographed image; a model storage part wherein a model expressing an object with a feature parameter is stored; a parameter value calculation part which calculates the value of the feature parameter in the model expressing the object photographed in the image in the feature area by adapting the image in the feature area in the photographed image to the model stored in the model storage part; and an output part which outputs the value of the feature parameter calculated by the parameter value calculation part, and an image in an area other than the feature area. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image processing system, an image processing method, and a program. The present invention particularly relates to an image processing system and an image processing method for processing an image, and a program for the image processing system.

A method of encoding and decoding a face image using a three-dimensional face model and eigenface decomposition is known (see, for example, Patent Document 1). In addition, the main image and a plurality of sub-images representing changes in the mouth portion in the main image are transmitted in advance, and then the main image is selected by selecting any of the plurality of sub-images in order to reproduce a moving image. There is known an image encoding device that transmits a code word that specifies whether or not to be combined (see, for example, Patent Document 2).
Japanese Patent Laid-Open No. 10-228544 Japanese Patent No. 2753599

  However, if the entire region of the image is encoded using a model, the amount of calculation increases. In addition, if it is possible to perform coding using a model up to a region having a low importance level, there is a possibility that the accuracy of reconstruction is reduced.

  In order to solve the above problem, according to a first aspect of the present invention, an image processing system stores a feature region detection unit that detects a feature region from a captured image, and a model in which an object is expressed by a feature parameter. The model storage unit that performs image processing and the feature region image in the captured image is adapted to the model stored in the model storage unit, thereby representing the feature parameter value in the model that represents the object captured in the feature region image. A parameter value calculation unit to calculate, and an output unit to output the value of the characteristic parameter calculated by the parameter value calculation unit and an image of an area other than the characteristic area.

  According to a second aspect of the present invention, there is provided an image processing method, a feature region detection stage for detecting a feature area from a captured image, a model storage stage for storing a model in which an object is expressed by a feature parameter, and a captured image A parameter value calculating step for calculating a value of a feature parameter in the model that represents an object captured in the image of the feature region by adapting the image of the feature region in the model stored in the model storing step; An output stage for outputting the feature parameter value calculated in the parameter value calculation stage and an image of an area other than the feature area.

  According to a third aspect of the present invention, there is provided a program for an image processing apparatus, wherein a computer stores a feature area detection unit that detects a feature area from a captured image, and a model storage that stores a model in which an object is represented by a feature parameter A parameter value for calculating a feature parameter value in the model representing an object imaged in the feature region image by adapting the image of the feature region in the captured image to the model stored in the model storage unit The calculation unit and the parameter value calculation unit function as an output unit that outputs the value of the characteristic parameter and an image of an area other than the characteristic area.

  The above summary of the invention does not enumerate all necessary features of the present invention. In addition, a sub-combination of these feature groups can also be an invention.

  Hereinafter, although this invention is demonstrated through embodiment of invention, the following embodiment does not limit the invention concerning a claim. In addition, not all the combinations of features described in the embodiments are essential for the solving means of the invention.

  FIG. 1 shows an example of an image processing system 10 according to an embodiment. The image processing system 10 can function as a monitoring system as described below.

  The image processing system 10 includes a plurality of imaging devices 100a to 100d (hereinafter collectively referred to as imaging devices 100) that capture the monitoring target space 150, an image processing device 120 that processes captured images captured by the imaging device 100, and communication. It includes a network 110, an image processing device 170, an image DB 175, and a plurality of display devices 180a-d (hereinafter collectively referred to as display devices 180). The image processing device 170 and the display device 180 are provided in a space 160 different from the monitoring target space 150.

  The imaging device 100a includes an imaging unit 102a and a captured image compression unit 104a. The imaging unit 102a captures a plurality of captured images by continuously capturing the monitoring target space 150. The captured image obtained by the imaging unit 102a may be a RAW format captured image. The captured image compression unit 104a synchronizes the RAW format captured image captured by the imaging unit 102a, compresses the captured moving image including a plurality of captured images obtained by the synchronization by MPEG encoding, and the like. Generate data. Thus, the imaging device 100a encodes the captured moving image obtained by capturing the monitoring target space 150 to generate captured moving image data. The imaging device 100a outputs the captured moving image data to the image processing device 120.

  Note that the imaging device 100b, the imaging device 100c, and the imaging device 100d have the same configuration as that of the imaging device 100a, and thus description of each component of the imaging device 100b, the imaging device 100c, and the imaging device 100d is omitted. In this way, the image processing device 120 acquires the captured moving image data generated by each of the plurality of imaging devices 100 from each of the plurality of imaging devices 100.

  Then, the image processing device 120 decodes the captured moving image data acquired from the imaging device 100 and acquires the captured moving image. The image processing apparatus 120 has different types of features such as an area where a person 130 is imaged, an area where a moving body 140 such as a vehicle is imaged, etc., from each of a plurality of captured images included in the acquired captured video. A plurality of feature regions are detected. Then, the image processing apparatus 120 compresses the image of the feature area with an intensity according to the type of the feature, and compresses the image of the area other than the feature area with an intensity stronger than the compression intensity for compressing the image of each feature area. You can do it.

  Note that the image processing apparatus 120 stores a mathematical model in which an object is represented by feature parameters. Then, the image processing apparatus 120 calculates the value of the feature parameter representing the image of the object by adapting the image of the object included in the feature region to the mathematical model.

  Then, the image processing device 120 generates feature area information including information for specifying the feature area detected from the captured image. Then, the image processing device 120 attaches the value of the feature parameter and the feature region information to the compressed moving image data and transmits the compressed moving image data to the image processing device 170 through the communication network 110.

  The image processing device 170 receives from the image processing device 120 compressed moving image data in which feature parameter values and feature region information are associated with each other. Then, the image processing apparatus 170 expands the received compressed moving image data using the associated feature area information. At this time, the image processing apparatus 170 generates an image of the object included in the feature area by changing the model with the value of the feature parameter, using the expanded image of the feature area. The display moving image generated in this way is supplied to the display device 180. The display device 180 displays the moving image for display supplied from the image processing device 170.

  Further, the image processing apparatus 170 may record the compressed moving image data and the characteristic parameter in the image DB 175 in association with the feature area information associated with the compressed moving image data. Then, in response to a request from the display device 180, the image processing device 170 reads the compressed moving image data, feature area information, and feature parameter values from the image DB 175, and generates a display moving image as described above. You may supply to the display apparatus 180. FIG.

  The feature area information is text data including the position of the feature area, the size of the feature area, the number of feature areas, identification information for identifying the captured image in which the feature area is detected, or the like, or compressed or encrypted into the text data. It may be data that has been processed. Then, the image processing apparatus 170 identifies captured images that satisfy various search conditions based on the position of the feature region, the size of the feature region, the number of feature regions, and the like included in the feature region information. Then, the image processing device 170 may decode the identified captured image and provide the decoded captured image to the display device 180.

  As described above, according to the image processing system 10, since the feature region is recorded in association with the moving image, it is possible to quickly search and find a picked-up image group that meets a predetermined condition in the moving image. Also, according to the image processing system 10, only a captured image group that meets a predetermined condition can be decoded, so that a partial moving image that meets the predetermined condition can be displayed promptly in response to a reproduction instruction.

  FIG. 2 shows an example of a block configuration of the image processing apparatus 120. The image processing device 120 includes an image acquisition unit 250, a feature region detection unit 203, a model storage unit 270, a parameter value calculation unit 260, a parameter quantization unit 280, a compression control unit 210, a compression unit 230, an association processing unit 206, and An output unit 207 is provided. The image acquisition unit 250 includes a compressed moving image acquisition unit 201 and a compressed moving image expansion unit 202.

  The compressed moving image acquisition unit 201 acquires a compressed moving image. Specifically, the compressed moving image acquisition unit 201 acquires encoded captured moving image data generated by the imaging device 100. The compressed moving image expansion unit 202 expands the captured moving image data acquired by the compressed moving image acquisition unit 201 and generates a plurality of captured images included in the captured moving image. Specifically, the compressed moving image decompression unit 202 decodes the encoded captured moving image data acquired by the compressed moving image acquisition unit 201, and generates a plurality of captured images included in the captured moving image. The captured image included in the captured moving image may be a frame image and a field image. The captured image in the present embodiment may be an example of a moving image constituent image in the present invention. In this way, the image acquisition unit 250 acquires a plurality of moving images captured by each of the plurality of imaging devices 100.

  The plurality of captured images obtained by the compressed moving image decompression unit 202 are supplied to the feature region detection unit 203 and the compression unit 230. The feature region detection unit 203 detects a feature region from a moving image including a plurality of captured images. Specifically, the feature region detection unit 203 detects a feature region from each of the plurality of captured images. Note that the captured moving image described above may be an example of a moving image in the following description.

  For example, the feature region detection unit 203 detects an image region whose image content changes in a moving image as a feature region. Specifically, the feature region detection unit 203 may detect an image region including a moving object as a feature region. Note that the feature region detection unit 203 may detect a plurality of feature regions having different types of features from each of the plurality of captured images. Note that the type of feature may be an index of the type of object such as a person and a moving object. The type of the object may be determined based on the degree of coincidence of the shape of the object or the color of the object. As described above, the feature region detection unit 203 may detect a plurality of feature regions having different types of included objects from a plurality of captured images.

  For example, the feature region detection unit 203 extracts an object that matches a predetermined shape pattern with a matching degree equal to or higher than a predetermined matching degree from each of the plurality of picked-up images, and in the picked-up image including the extracted object. The region may be detected as a feature region having the same feature type. Note that a plurality of shape patterns may be determined for each type of feature. As an example of the shape pattern, a shape pattern of a human face can be exemplified. Different face patterns may be determined for each of a plurality of persons. Thereby, the feature region detection unit 203 can detect different regions each including a different person as different feature regions. The feature region detection unit 203 includes, in addition to the person's face, a part of the human body such as the head of the person or the hand of the person, or at least a part of a living body other than the human body. Can be detected as a feature region. The living body includes a specific tissue existing inside the living body such as a tumor tissue or a blood vessel inside the living body. In addition to the living body, the feature region detection unit 203 may detect, as the feature region, a region in which a card such as money, a cash card, a vehicle, or a license plate of the vehicle is captured.

  In addition to pattern matching by template matching or the like, the feature region detection unit 203 detects feature regions based on learning results by machine learning (for example, Adaboost) described in, for example, Japanese Patent Application Laid-Open No. 2007-188419. It can also be detected. For example, an image feature amount extracted from a predetermined subject image and an image feature amount extracted from a subject image other than the predetermined subject are extracted from the predetermined subject image. Learn the features of the image features. Then, the feature region detection unit 203 may detect a region from which an image feature amount having a feature that matches the learned feature is extracted as a feature region. Thereby, the feature region detection unit 203 can detect a region where a predetermined subject is imaged as a feature region.

  As described above, the feature region detection unit 203 detects a plurality of feature regions from a plurality of captured images included in each of the plurality of moving images. Then, the feature region detection unit 203 supplies information indicating the detected feature region to the compression control unit 210. Note that the information indicating the feature region includes coordinate information of the feature region indicating the position of the feature region, type information indicating the type of the feature region, and information for identifying the captured moving image in which the feature region is detected. As described above, the feature region detection unit 203 detects the feature region in the moving image.

  The compression control unit 210 controls the compression processing of the moving image by the compression unit 230 according to the feature region based on the information indicating the feature region acquired from the feature region detection unit 203. Note that the compression unit 230 may compress the captured image by compressing the feature region in the captured image and the region other than the feature region in the captured image with different intensities. For example, the compression unit 230 compresses the captured image by reducing the resolution of the region other than the feature region in the captured image included in the moving image. As described above, the compression unit 230 compresses the image in the region other than the feature region by reducing the image quality of the image in the region other than the feature region. The compression unit 230 compresses each image region in the captured image with an intensity corresponding to the importance. A more specific internal compression operation of the compression unit 230 will be described later.

  The model storage unit 270 stores a model in which an object is expressed by feature parameters. For example, the model storage unit 270 may store a model in which an object is expressed by statistical feature parameters. More specifically, the model storage unit 270 may store a model in which an object is expressed with principal components based on principal component analysis. Note that the model storage unit 270 may store a model in which the shape of an object is expressed using principal components based on principal component analysis. In addition, the model storage unit 270 may store a model in which the color of an object is expressed using principal components based on principal component analysis.

  The parameter value calculation unit 260 adapts the image of the object included in the feature region image in the captured image to the model stored in the model storage unit 270, thereby representing the object included in the feature region image. The value of the feature parameter at is calculated. Specifically, the parameter value calculation unit 260 calculates the weight of the principal component in the model. When the feature parameter is a principal component vector obtained by principal component analysis, a weighting coefficient for the principal component vector can be exemplified as the value of the feature parameter.

  The parameter quantization unit 280 selects the feature parameter for which the value of the feature parameter is to be output from the output unit 207. Specifically, parameter quantization section 280 determines which level of principal components extracted from the principal components analysis should be output with weighting coefficients. For example, the parameter quantization unit 280 determines that the weighting coefficient of the principal component should be output to a predetermined level according to the feature type of the feature region. The weighting coefficient for the principal component at the level determined by the parameter quantization unit 280 is supplied to the association processing unit 206.

  The association processing unit 206 associates information specifying a feature region detected from the captured image and a weighting coefficient with the captured image. Specifically, the association processing unit 206 associates information specifying a feature region detected from a captured image and a weighting coefficient with a compressed moving image including the captured image as a moving image constituent image. Then, the output unit 207 outputs to the image processing apparatus 170 the compressed moving image in which the information specifying the feature region and the weighting coefficient are associated by the association processing unit 206.

  As described above, the output unit 207 outputs the feature parameter value calculated by the parameter value calculation unit 260 and an image of a region other than the feature region. More specifically, the output unit 207 outputs the value of the feature parameter selected by the parameter quantization unit 280 and an image of a region other than the feature region whose image quality has been reduced by the compression unit 230.

  Note that the compressed moving image output from the output unit 207 may not include pixel information for the feature region. In this manner, the output unit 207 outputs the image of the region other than the principal component weight and the feature region calculated by the parameter value calculation unit 260. More specifically, the output unit 207 outputs the feature parameter value calculated by the parameter value calculation unit 260 and an image of a region other than the feature region whose image quality has been reduced by the compression unit 230.

  As described above, according to the image processing apparatus 120, the object image included in the feature region is represented by the feature parameter value, thereby reconstructing the object image later while sufficiently reducing the data amount. You can leave information that you can. In addition, for the background area that is less important than the feature area, the amount of data can be significantly reduced by reducing the image quality.

  The model storage unit 270 may store models of different types of objects in association with the types. Then, the parameter value calculation unit 260 adapts the image of the object included in the image of the feature region in the captured image to the model stored in the model storage unit 270 in association with the type of the object captured in the feature region. By doing so, the value of the characteristic parameter may be calculated. In this case, the output unit 207 displays the value of the feature parameter calculated by the parameter value calculation unit 260, the type of the object captured in the feature region, and the image of the region other than the feature region whose image quality has been reduced by the compression unit 230. It is desirable to output. Accordingly, an appropriate type of model can be selected and reconfigured on the image processing apparatus 170 side.

  In addition, the model storage unit 270 may store an object model viewed from a different direction in association with the direction. The parameter value calculation unit 260 stores the image of the object included in the image of the feature area in the captured image in the model storage unit 270 in association with the direction in which the object imaged in the feature area is captured. By adapting to the model, the value of the feature parameter may be calculated. In this case, the output unit 207 outputs the value of the feature parameter calculated by the parameter value calculation unit 260, the direction in which the object imaged in the feature region is captured, and the region other than the feature region whose image quality has been reduced by the compression unit 230 It is desirable to output the image.

  The model storage unit 270 may store models of objects illuminated under different illumination conditions in association with the illumination conditions. The parameter value calculation unit 260 stores the image of the object included in the image of the feature area in the captured image in association with the illumination condition in which the object imaged in the feature area is illuminated. The value of the feature parameter may be calculated by adapting to the model. In this case, the output unit 207 includes the value of the feature parameter calculated by the parameter value calculation unit 260, the illumination condition in which the object imaged in the feature region is illuminated, and the feature region whose image quality is reduced by the compression unit 230 It is desirable to output an image of the area.

  As described above, the model storage unit 270 stores a plurality of models in association with the type of object, the direction of the object, the illumination condition, and the like. For this reason, since the image of the object in the feature region can be expressed using a more appropriate model, the amount of data can be reduced while maintaining the image quality of the feature region.

  FIG. 3 shows an example of a block configuration of the compression unit 230. The compression unit 230 includes an image division unit 232, a plurality of fixed value conversion units 234a-c (hereinafter, may be collectively referred to as a fixed value conversion unit 234), and a plurality of image quality conversion units 241a-d (hereinafter, image quality conversion units). And a plurality of compression processing units 236a-d (hereinafter may be collectively referred to as compression processing units 236).

  The image dividing unit 232 acquires a plurality of captured images from the image acquisition unit 250. Then, the image dividing unit 232 divides the plurality of captured images into a feature region and a background region other than the feature region. Specifically, the image dividing unit 232 divides the plurality of captured images into each of a plurality of feature areas and a background area other than the feature areas. As described above, the image dividing unit 232 divides each of the plurality of captured images into the feature region and the background region.

  Then, the compression processing unit 236 compresses the feature region image, which is the feature region image, and the background region image, which is the background region image, with different strengths. Specifically, the compression processing unit 236 compresses a feature area moving image including a plurality of feature area images and a background area moving image including a plurality of background area images with different strengths.

  Specifically, the image dividing unit 232 generates a feature area moving image for each of a plurality of feature types by dividing a plurality of captured images. Then, the fixed value unit 234 fixes the pixel values of the regions other than the feature regions of the respective feature types for each of the feature region images included in the plurality of feature region moving images generated for each feature type. To do. Specifically, the fixed value converting unit 234 sets pixel values in regions other than the feature region to predetermined pixel values.

  The image quality conversion unit 241 converts the image quality of the feature area image and the background area image. For example, the image quality conversion unit 241 converts at least one of the resolution, the number of gradations, the dynamic range, and the number of included colors for each of the divided feature region image and background region image. Then, the compression processing unit 236 compresses a plurality of feature area moving images for each feature type. For example, the compression processing unit 236 performs MPEG compression on a plurality of feature area moving images for each feature type.

  Note that the fixed value unit 234a, the fixed value unit 234b, and the fixed value unit 234c are a feature region moving image of the first feature type, a feature region moving image of the second feature type, and a third feature, respectively. The type of feature area animation is fixed. In addition, the image quality conversion unit 241a, the image quality conversion unit 241b, the image quality conversion unit 241c, and the image quality conversion unit 241d respectively include a feature area moving image of the first feature type, a feature area moving image of the second feature type, and a third The image quality of the feature region moving image of the feature type and the background region moving image is converted. Then, the compression processing unit 236a, the compression processing unit 236b, the compression processing unit 236c, and the compression processing unit 236d are a feature region moving image of the first feature type, a feature region moving image of the second feature type, and a third feature. Types of feature area animation and background area animation are compressed.

  Note that the compression processing units 236a-c compress the feature area moving image with a predetermined strength according to the type of feature. For example, the compression processing unit 236 may convert the feature area moving image to a different resolution determined in advance according to the type of the feature, and compress the converted feature area moving image. In addition, when compressing the feature region moving image by MPEG encoding, the compression processing unit 236 may compress the feature region moving image with different quantization parameters determined in advance according to the feature type.

  The compression processing unit 236d compresses the background area moving image. Note that the compression processing unit 236d may compress the background region moving image with a strength higher than the strength of any of the compression processing units 236a-c. The feature area moving image and the background area moving image compressed by the compression processing unit 236 are supplied to the association processing unit 206.

  Since regions other than the feature region have been fixed values by the fixed value unit 234, when the compression processing unit 236 performs predictive encoding by MPEG encoding or the like, the region other than the feature region may be connected to the predicted image. The amount of difference between the images can be significantly reduced. Therefore, the compression rate of the feature area moving image can be significantly increased.

  In this way, the compression unit 230 generates an image that becomes an input image to the image processing device 170 by reducing the image quality of the captured image. Specifically, the compression unit 230 generates an image to be an input image to the image processing apparatus 170 by reducing the resolution, the number of gradations, and the number of colors used in the captured image. For example, the compression unit 230 may generate an image to be an input image to the image processing device 170 by further reducing higher spatial frequency components in the captured image.

  In this figure, each of the plurality of compression processing units 236 included in the compression unit 230 compresses a plurality of feature region images and a background region image. However, in another embodiment, the compression unit 230 performs one compression process. A single compression processing unit 236 may compress a plurality of feature region images and background region images with different intensities. For example, a plurality of feature region images and a background region image are sequentially supplied to one compression processing unit 236 in a time-sharing manner, and the one compression processing unit 236 differs from the plurality of feature region images and the background region image. You may compress sequentially by intensity.

  In addition, the one compression processing unit 236 quantizes the image information of the plurality of feature regions and the image information of the background region with different quantization coefficients, respectively, thereby converting the images of the plurality of feature regions and the images of the background region. They may be compressed with different strengths. Also, an image obtained by converting a plurality of feature region images and a background region image into images of different image quality is supplied to one compression processing unit 236, and the one compression processing unit 236 includes a plurality of feature region images and Each image in the background area may be compressed. The image quality conversion process may be performed by one image quality conversion unit 240. Further, as described above, one compression processing unit 236 quantizes with a different quantization coefficient for each region, or one compression processing unit 236 compresses an image converted into a different image quality for each region. The compression processing unit 236 may compress one image, or may compress each of the images divided by the image dividing unit 232 as described with reference to FIG. When one compression processing unit 236 compresses one image, the image dividing unit 232 does not need to perform the dividing process and the fixed value converting unit 234 does not perform the fixing process. The dividing unit 232 and the fixed value unit 234 may not be provided.

  FIG. 4 shows an example of a block configuration of the image processing apparatus 170. The image processing apparatus 170 includes an image acquisition unit 301, an association analysis unit 302, an expansion control unit 310, an expansion unit 320, an image generation unit 380, a feature area information acquisition unit 360, a model storage unit 350, and an output unit 340. The image generation unit 380 includes an enlargement unit 332 and a synthesis unit 330.

  The image acquisition unit 301 acquires the compressed moving image compressed by the compression unit 230. Specifically, the image acquisition unit 301 acquires a compressed moving image including a plurality of feature area moving images and a background area moving image. More specifically, the image acquisition unit 301 acquires a compressed moving image with feature area information and feature parameters. In this way, the image acquisition unit 301 acquires a captured image with reduced feature parameter values and image quality output from the output unit 207. In particular, the image acquisition unit 301 acquires a captured image with reduced image quality and feature parameter values, particularly in a region other than the feature region.

  Then, the association analysis unit 302 separates the moving image data acquired by the image acquisition unit 301 into a plurality of feature region moving images and background region moving images, feature region information, and feature region parameter values, and a plurality of feature region moving images. The background area moving image is supplied to the decompression unit 320. Further, the association analysis unit 302 supplies the position of the feature region and the feature type to the extension control unit 310 and the feature region information acquisition unit 360. In addition, the association analysis unit 302 supplies the feature parameter value to the feature region information acquisition unit 360. In this manner, the feature region information acquisition unit 360 can acquire information indicating the feature region in each of the plurality of captured images (for example, information indicating the position of the feature region) and the feature parameter value. The feature region information acquisition unit 360 supplies information indicating the position of the acquired feature region and the feature parameter value to the image generation unit 380.

  The decompression control unit 310 controls the decompression processing by the decompression unit 320 according to the position of the feature region and the feature type acquired from the association analysis unit 302. For example, the decompression control unit 310 decompresses each region of the moving image indicated by the compressed moving image to the decompressing unit 320 according to the compression method in which the compression unit 230 compresses each region of the moving image according to the position of the feature region and the type of feature. Let

  Hereinafter, the operation of each component included in the extension unit 320 will be described. The decompression unit 320 includes a plurality of decoders 322a-d (hereinafter collectively referred to as decoders 322). The decoder 322 decodes one of the plurality of encoded characteristic area moving images and background area moving images. Specifically, the decoder 322a, the decoder 322b, the decoder 322c, and the decoder 322d decode the first feature region moving image, the second feature region moving image, the third feature region moving image, and the background region moving image, respectively. The decompressing unit 320 supplies the first feature region moving image, the second feature region moving image, the third feature region moving image, and the background region moving image obtained by decoding to the image generating unit 380.

  The image generation unit 380 generates one display moving image based on the first feature region moving image, the second feature region moving image, the third feature region moving, the background region moving image, and the feature region information. Then, the output unit 340 outputs the feature area information and the display moving image acquired from the association analysis unit 302 to the display device 180 or the image DB 175. The image DB 175 associates the position of the feature region indicated by the feature region information, the feature type of the feature region, and the number of feature regions with the information for identifying the captured image included in the display moving image, and the non-volatile memory such as a hard disk. May be recorded on the recording medium. The output unit 340 can function as an image output unit in the present invention.

  The model storage unit 350 stores the same model as the model stored in the model storage unit 270. Then, the image generation unit 380 may generate a high-quality image of the object included in the feature region by adapting the image of the object included in the feature region to the model stored in the model storage unit 350. Specifically, the image generation unit 380 generates a high-quality image of the object by weighting the principal component vector stored in the model storage unit 350 with a weighting coefficient as an example of the feature parameter value. It's okay. As described above, the image generation unit 380 generates an image of an object included in the image of the feature region from the model based on the value of the feature parameter.

  Note that the parameter value calculation unit 260 adapts the image of the object included in the feature region image in the captured image to the model stored in the model storage unit 270, thereby capturing the object captured in the feature region image. The value of the feature parameter in the model that represents the shape of the model may be calculated. Then, the compression unit 230 may compress the captured image by reducing the image quality of the feature region and the region other than the feature region in the captured image. The output unit 207 may output the characteristic parameter value calculated by the parameter value calculation unit 260 and the captured image whose image quality has been reduced by the compression unit 230.

  In this case, the image generation unit 380 generates the shape of the object captured in the image of the feature region from the model based on the value of the feature parameter, and the generated object shape and the captured image acquired by the image acquisition unit 250 An image of an object captured in the feature region image is generated using the pixel value of the feature region image. Specifically, the image generation unit 380 generates the shape of the object captured in the image of the feature region from the model based on the value of the feature parameter, and the generated object shape and the feature expanded by the expansion unit 320 An image of the object captured in the feature region image is generated using the pixel value of the region image.

  Note that the feature region information acquisition unit 360 may acquire the type of object, the direction of the object, and the illumination condition output from the output unit 207 in association with the compressed moving image. Then, the image generation unit 380 uses the weighting coefficient acquired by the feature region information acquisition unit 360 to store the principal component vector stored in the model storage unit 350 in association with the object type, the object direction, and the illumination condition. A high quality image of the object may be generated by weighting.

  The image enlargement unit 332 enlarges an image in an area other than the feature area. Then, the synthesis unit 330 synthesizes a high-quality image of the object in the feature area and an image of the area other than the enlarged feature area.

  Then, the output unit 340 outputs an image including a high-quality image and an image other than the feature region. Specifically, the output unit 340 outputs a display moving image including the captured image obtained by the combining unit 330 as a moving image constituent image as described above.

  FIG. 5 shows an example of another block configuration of the compression unit 230. The compression unit 230 in this configuration compresses a plurality of captured images by a spatial scalable encoding process corresponding to the type of feature.

  The compression unit 230 in this configuration includes an image quality conversion unit 510, a difference processing unit 520, and an encoding unit 530. The difference processing unit 520 includes a plurality of inter-layer difference processing units 522a-d (hereinafter collectively referred to as inter-layer difference processing units 522). Encoding section 530 includes a plurality of encoders 532a-d (hereinafter collectively referred to as encoders 532).

  The image quality conversion unit 510 acquires a plurality of captured images from the image acquisition unit 250. In addition, the image quality conversion unit 510 acquires information specifying the feature region detected by the feature region detection unit 203 and information specifying the type of feature in the feature region. Then, the image quality conversion unit 510 duplicates the captured image, and generates captured images of the number of types of features in the feature area. Then, the image quality conversion unit 510 converts the generated captured image into an image having a resolution corresponding to the type of feature.

  For example, the image quality conversion unit 510 has a captured image converted to a resolution corresponding to the background area (hereinafter referred to as a low resolution image), and a captured image converted to the first resolution corresponding to the type of the first feature ( Called a first resolution image), a captured image converted to a second resolution corresponding to the second feature type (referred to as a second resolution image), and a third resolution corresponding to the third feature type. A captured image (referred to as a third resolution image) converted to is generated. Here, it is assumed that the first resolution image has a higher resolution than the low resolution image, the second resolution image has a higher resolution than the first resolution image, and the third resolution image has a higher resolution than the second resolution image.

  Then, the image quality conversion unit 510 converts the low resolution image, the first resolution image, the second resolution image, and the third resolution image into the inter-layer difference processing unit 522d, the inter-layer difference processing unit 522a, and the inter-layer difference processing unit 522b, respectively. And to the inter-layer difference processing unit 522c. The image quality conversion unit 510 supplies a moving image to each of the inter-layer difference processing units 522 by performing the above-described image quality conversion processing on each of the plurality of captured images.

  Note that the image quality conversion unit 510 may convert the frame rate of the moving image supplied to each of the inter-layer difference processing unit 522 in accordance with the feature type of the feature region. For example, the image quality conversion unit 510 may supply, to the inter-layer difference processing unit 522d, a moving image having a lower frame rate than the moving image supplied to the inter-layer difference processing unit 522a. In addition, the image quality conversion unit 510 may supply a moving image having a lower frame rate than the moving image supplied to the inter-layer difference processing unit 522b to the inter-layer difference processing unit 522a, and a frame lower than the moving image supplied to the inter-layer difference processing unit 522c. The rate movie may be supplied to the inter-tier difference processing unit 522b. Note that the image quality conversion unit 510 may convert the frame rate of the moving image supplied to the inter-layer difference processing unit 522 by thinning out the captured image according to the feature type of the feature region. Note that the image quality conversion unit 510 can perform the same image quality conversion processing as the image quality conversion unit 241 described with reference to FIG.

  The inter-layer difference processing unit 522d and the encoder 532d predictively encode a background area moving image including a plurality of low-resolution images. Specifically, the inter-layer difference processing unit 522 generates a difference image from a predicted image generated from another low-resolution image. Then, the encoder 532d quantizes the transform coefficient obtained by converting the difference image into a spatial frequency component, and encodes the quantized transform coefficient by entropy coding or the like. Note that such predictive encoding processing may be performed for each partial region of the low-resolution image.

  Further, the inter-layer difference processing unit 522a predictively encodes the first feature region moving image including the plurality of first resolution images supplied from the image quality conversion unit 510. Similarly, the inter-layer difference processing unit 522b and the inter-layer difference processing unit 522c each predictively encode a second feature area moving image including a plurality of second resolution images and a third feature area moving image including a plurality of third resolution images. To do. Hereinafter, specific operations of the inter-layer difference processing unit 522a and the encoder 532a will be described.

  The inter-layer difference processing unit 522a decodes the first resolution image encoded by the encoder 532d, and expands the decoded image to an image having the same resolution as the first resolution. Then, the inter-layer difference processing unit 522a generates a difference image between the enlarged image and the low resolution image. At this time, the inter-layer difference processing unit 522a sets the difference value in the background area to zero. Then, the encoder 532a encodes the difference image in the same manner as the encoder 532d. Note that the encoding processing by the inter-layer difference processing unit 522a and the encoder 532a may be performed for each partial region of the first resolution image.

  Note that, when the first resolution image is encoded, the inter-layer difference processing unit 522a calculates the code amount predicted when the difference image with the low resolution image is encoded, and the other first resolution image. The amount of code predicted when the difference image between the generated prediction image and the prediction image is encoded is compared. In the case where the latter code amount is smaller, the inter-layer difference processing unit 522a generates a difference image from the predicted image generated from the other first resolution image. In addition, the inter-layer difference processing unit 522a, when it is predicted that the code amount is smaller when encoded without taking the difference from the low-resolution image or the predicted image, It is not necessary to take a difference between them.

  Note that the inter-layer difference processing unit 522a does not have to set the difference value in the background area to zero. In this case, the encoder 532a may set the encoded data for difference information in an area other than the feature area to zero. For example, the encoder 532a may set the conversion coefficient after conversion to a frequency component to zero. Note that the motion vector information when the inter-layer difference processing unit 522d performs predictive encoding is supplied to the inter-layer difference processing unit 522a. The inter-layer difference processing unit 522a may calculate a motion vector for a predicted image using the motion vector information supplied from the inter-layer difference processing unit 522d.

  Note that the operations of the inter-layer difference processing unit 522b and the encoder 532b are that the second resolution image is encoded, and the first resolution after encoding by the encoder 532a when the second resolution image is encoded. The operations of the inter-layer difference processing unit 522b and the encoder 532b are substantially the same as the operations of the inter-layer difference processing unit 522a and the encoder 532a, except that the difference from the image may be taken. To do. Similarly, the operations of the inter-layer difference processing unit 522c and the encoder 532c are that the third resolution image is encoded, and when the third resolution image is encoded, after the encoding by the encoder 532b. Except for the fact that a difference from the second resolution image may be obtained, the operation is substantially the same as the operation of the inter-layer difference processing unit 522a and the encoder 532a, and thus the description thereof is omitted.

  As described above, the image quality conversion unit 510 generates, from each of the plurality of captured images, a low image quality image with a low image quality and a feature region image with higher image quality than the low image quality at least in the feature region. Then, the difference processing unit 520 generates a feature region difference image indicating a difference image between the feature region image in the feature region image and the feature region image in the low-quality image. Then, the encoding unit 530 encodes the feature region difference image and the low quality image.

  In addition, the image quality conversion unit 510 generates a low-quality image with reduced resolution from a plurality of captured images, and the difference processing unit 520 includes a feature region image in the feature region image and a feature region image in the low-quality image. A feature region difference image between the image and the image enlarged is generated. Further, the difference processing unit 520 has a spatial frequency component in which the difference between the feature region image and the enlarged image in the feature region is converted into the spatial frequency region, and the data amount of the spatial frequency component is in the region other than the feature region. A reduced feature area difference image is generated.

  As described above, the compression unit 230 performs hierarchical encoding by encoding image differences between a plurality of layers having different resolutions. As is clear from this, a part of the compression method by the compression unit 230 of this configuration is H.264. It is clear that a compression scheme according to H.264 / SVC is included. Note that when the image processing apparatus 170 decompresses such a layered compressed moving image, the moving image data of each layer is decoded, and the difference is taken for the region encoded by the inter-layer difference. The captured image having the original resolution can be generated by the addition process with the captured image decoded in the hierarchy.

  FIG. 6 shows an example of feature points on a human face. As described with reference to FIGS. 2 and 3, the model storage unit 270 and the model storage unit 350 store models that express objects with feature parameters. Hereinafter, as an example of a generation method for generating models stored in the model storage unit 270 and the model storage unit 350, a case where a model of a human face, which is an example of an object, is generated using an AAM method will be described.

  As shown in FIG. 8, n feature points representing the face shape are set for each of a plurality of face images (hereinafter referred to as sample images) representing a human face portion as a sample. Here, it is assumed that the number of feature points is smaller than the number of pixels of the face image. For each feature point, which part of the face is shown may be determined in advance, for example, the first feature point is the left end of the left eye and the eleventh feature point is the center between the eyebrows. Each feature point may be set manually or automatically by a recognition process.

Then, the average shape of the face is calculated based on the feature points set in each sample image. Specifically, the average of the position coordinates for each feature point indicating the same part in each sample image is obtained. Then, principal component analysis is performed based on the feature points representing the face shape in each sample image and the position coordinates of the average shape. As a result, an arbitrary face shape S can be expressed by S = S ₀ + Σp _i b _i (i = 1 to n).

Here, S is the shape vector is represented by arranging the position coordinates of each feature point of the face shape (x1, y1, ···, x n, y n) is, S ₀ is of the feature points in the average face shape the average face shape vectors represented side by side position coordinates, p _i is the eigenvector representing the i-th principal component of the face shape obtained by the principal component analysis, b _i represents a weighting factor for each eigenvector p _i.

FIG. 7 schematically shows an example of a change in face shape when the weighting coefficient b is changed. In the figure, how the face shape changes when the values of the weighting coefficients b ₁ and b ₂ for the eigenvectors p ₁ and p ₂ of the top two principal components obtained by the principal component analysis are changed. Represents. a) shows the change in the face shape when changing the weighting coefficients b _1, b) shows a change of face shape when changing the weighting coefficients b _2. In both a) and b), the central one of the three face shapes for each principal component shows an average face shape.

In this example, as a result of the principal component analysis, a component contributing to the contour shape of the face is extracted as the first principal component. By changing the weighting coefficient b ₁ , a) from the elongated face shown at the left end a ) It can be seen that the face shape has changed to the round face shown at the right end. Similarly, the second principal component are extracted component contributing to the length of the open and closed states and jaw of the mouth, by varying the weighting factor b _2, b) mouth, as shown at the left end open It can be seen that the face shape changes from a face with a long chin in the state to a face with a short chin with the mouth closed as shown at the right end. It should be noted that what shape element each component contributes to depends on human interpretation. Through the principal component analysis, a sample representing a larger shape difference in each used sample image is extracted as a lower-order principal component.

  FIG. 8 shows an example of an image obtained by converting a sample image into an average face shape. Each sample image is converted (warped) into an average face shape. Specifically, the shift amount between each sample image and the average face shape is calculated for each feature point. Then, based on the shift amount, the shift amount to the average face shape for each pixel of each sample image is calculated, and each sample image is warped to the average face shape for each pixel.

Then, principal component analysis is performed using the pixel values of the R, G, and B color components of each pixel for each sample image after conversion to the average face shape as variables. As a result, the pixel values of the R, G, and B color components under the average face shape of an arbitrary face image can be approximated by A = A ₀ + Σq _i λ _i (i = 1 to m).

  Here, A is a vector (r1, g1, b1, r2, g2, b2,..., Which is expressed by arranging the pixel values of the R, G, and B color components of each pixel under the average face shape. rm, gm, bm). Note that r, g, and b are pixel values of R, G, and B color components, 1 to m are subscripts for identifying each pixel, and m is the total number of pixels in the average face shape. The arrangement order of the vector components is not limited to the above order.

A ₀ is an average vector expressed by arranging average values of pixel values of R, G, and B color components for each pixel of each sample image in the average face shape, and q _i is a face obtained by principal component analysis. The eigenvector representing the i-th principal component for the pixel values of the R, G, B color components, λ _i represents a weighting coefficient for each eigenvector q _i .

FIG. 9 schematically shows an example of a change in pixel value when the weighting coefficient q is changed. This figure schematically shows how the face pixel values change when the values of the weighting coefficients λ ₁ and λ ₂ for the eigenvectors q ₁ and q ₂ of the top two principal components obtained by principal component analysis are changed. It expresses. a) shows the change of the pixel value when the weighting coefficient λ ₁ is changed, and b) shows the change of the pixel value when the weighting coefficient λ ₂ is changed. In both a) and b), the center of the three face shapes for each principal component indicates an average pixel value.

In this example, the results of principal component analysis, the first principal component are extracted component contributing to the presence or absence of beard, by varying the weighting factor lambda _1, from the face without beard shown in a) left a) It can be seen that the face changes to the dark face shown at the right end. A component contributing to the darkness of the eyebrows is extracted as the second main component, and by changing the weighting coefficient λ ₂ , b) the eyebrows are thin as shown at the left end, and b) are shown at the right end. It can be seen that the navel changes to a dark face.

As described above, the face model is generated by the processing described with reference to FIGS. This model represents a face by a plurality of eigenvectors p _i representing the face shape and eigenvectors q _i representing the pixel values of the face under the average face shape. The total number of eigenvectors of the model is significantly smaller than the number of pixels forming the face image. In the above example, the shape and pixel value are expressed by separate weighting coefficients b _i and λ _i for the face shape and the pixel values of the R, G, and B color components. Since there is a correlation between pixel value variations of color components and color components, it is possible to perform principal component analysis on feature parameters including both feature points and pixel values.

  Next, an example of processing for compressing an image of an object included in a feature region using a model stored in the model storage unit 270 will be described. The parameter value calculation unit 260 normalizes the input face image included in the feature region, and calculates the pixel values of the R, G, and B color components in the average face shaping. Note that the input face image may not necessarily be an image from the front, or the illumination condition may be different from the illumination condition when a sample image is captured. Therefore, normalization here refers not only to the process of aligning the feature points of the front face described above, but also to converting the orientation of the input face image obtained by imaging from the diagonal to the face image captured from the front. And conversion processing into a face image captured in a shooting environment equivalent to that of the sample image, such as shadow removal processing for removing the influence of shadows due to illumination.

Then, the parameter value calculation unit 260 calculates the weighting coefficient λ _i by projecting the difference from the pixel value for the average face onto the principal component vector q _i . Specifically, the parameter value calculation unit 260 can calculate the weighting coefficient λ _i by the inner product with the principal component vector q _i . The parameter value calculation unit 260 also calculates the facial feature point S by the same process as the pixel value A calculation process described above. Specifically, the parameter value calculation unit 260 calculates the weighting coefficient b _i by projecting the difference in the position of the feature point with respect to the average face onto the principal component vector p _i .

As described above, the parameter value calculation unit 260 can calculate the weighting coefficients b _i and λ _i as characteristic parameter values. Next, high-quality image generation processing by the image generation unit 380 will be described.

The image generation unit 380 calculates the pixel value A under the average face shape using the acquired weighting coefficient λ _i , average face pixel value A ₀ , and principal component vector q _i . In addition, the image generation unit 380 calculates the feature point A using the acquired weighting coefficient b _i , average face feature point S ₀ , and principal component vector p _i . Then, the image generation unit 380 performs an inverse conversion process on the image represented by the pixel value A and the feature point A except for the process of aligning the feature points in the normalization process. Note that the content of the normalization process may be transmitted from the image processing apparatus 120 to the image processing apparatus 170 and used when the image generation unit 380 performs the inverse conversion process.

  Through the above processing, the image generation unit 380 generates a high-quality image with higher image quality than the captured image from the image of the feature region in the captured image output from the output unit 207. Specifically, the image generation unit 380 has a higher resolution image, a sharper image, an image with less noise, and a greater number of gradations than the image of the feature region in the captured image output from the output unit 207. It is possible to generate an image with a large number of images or an image with a larger number of colors.

  FIG. 10 shows an example of a model stored in the model storage unit 270 and the model storage unit 350 in a table format. The model storage unit 270 and the model storage unit 350 store models for each combination of a plurality of facial expressions and a plurality of directions. Expressions include faces and true faces in each state of emotions, and directions include front, upper, lower, right, left, and rear.

  The parameter value calculation unit 260 identifies the facial expression and the face direction based on the image content of the face image included in the feature region, and the model storage unit 270 stores the facial expression and the direction in association with the identified facial expression and direction. The above-described weighting coefficient can be calculated using the model that is used.

  Note that the output unit 207 may transmit information for identifying the used model to the image processing apparatus 170. Then, the image generation unit 380 can perform the above-described reconstruction process using the model identified by the information.

  The image generation unit 380 can specify a facial expression from the shape of the mouth and / or eyes, and can specify the direction of the face from the positional relationship of the eyes, mouth, nose, and ears. The image processing apparatus 120 may specify the facial expression and the face direction, and the output unit 207 may output the facial expression and the face direction in association with the captured image.

  In addition to the facial expression and orientation, the model storage unit 270 and the model storage unit 350 may store models in association with illumination conditions as described above. For example, the model storage unit 270 and the model storage unit 350 may store models in association with the illumination intensity and the illumination direction. Then, the parameter value calculation unit 260 may specify the illumination condition for the face based on the image content of the face image included in the feature region. For example, the parameter value calculation unit 260 identifies the illumination intensity and illumination direction based on the position and size of the shadow, and stores the model stored in the model storage unit 270 in association with the identified illumination intensity and illumination direction. Can be used to calculate the weighting factor.

  In the above example, the generation of a model that represents the entire face and the feature parameter extraction and reconstruction process using the model have been described. In addition to such a model of the entire face, the image processing system 10 can use a model for each part of the face. In addition, the image processing system 10 can use a model of a face (or a part of a face) for each gender and / or race. The image processing system 10 can store a model for each type of object that the image processing system 10 is to monitor, such as a vehicle and a ship, in addition to a human model. The image generation unit 380 can also select and reconfigure a model according to the type. Such a type may be detected by the image processing apparatus 120 and transmitted to the image processing apparatus 170 in association with the captured image.

  As described above, the model storage unit 270 and the model storage unit 350 can store models of different types of objects in association with the types. Then, the feature area information acquisition unit 360 acquires information indicating the type of object imaged in the feature area in the input image. Then, the image generation unit 380 stores the image of the object included in the feature region in the captured image in association with the type of the object captured in the feature region acquired by the feature region information acquisition unit 360, and is stored in the model storage unit 350. The image is converted into a high-quality image by adapting to the model.

  Note that, as described above, the model storage unit 270 and the model storage unit 350 convert a model as an example of learning data into a facial part (eg, eyes, nose, mouth, etc.) as an example of information identifying the type of object. ) Can be stored every time. Here, the learning data may include a low frequency component and a high frequency component of the image of the object respectively extracted from a large number of sample images of the object in addition to the model. Note that the low frequency components of the object image are clustered into a plurality of clusters in each of the plurality of object types by clustering the low frequency components of the object image for each of the plurality of object types by the K-means method or the like. May have been. Moreover, a typical low frequency component (for example, a centroid value) may be determined for each cluster. Note that the model storage unit 270 may store information for specifying the high frequency component in association with the low frequency component of the image of the object. Further, the model storage unit 350 may store the high frequency component in association with information specifying the high frequency component.

  The parameter value calculation unit 260 extracts a low frequency component from the image of the object included in the captured image. Then, the parameter value calculation unit 260 determines a value suitable for the extracted low frequency component as a representative low frequency component from the low frequency component cluster extracted from the sample image of the object type object. Identify the cluster. Then, the parameter value calculation unit 260 identifies information for identifying the high frequency component cluster stored in the model storage unit 270 in association with the low frequency component included in the identified cluster. In this way, the parameter value calculation unit 260 can specify a cluster of high frequency components that are correlated with the low frequency components extracted from the objects included in the captured image. Information identifying the high-frequency component cluster identified by the parameter value calculation unit 260 is output from the output unit 207 in association with the information identifying the feature region.

  Information identifying the high-frequency component clusters output from the output unit 207 and acquired by the image acquisition unit 301 is extracted by the association analysis unit 302 and supplied to the image generation unit 380 through the feature region information acquisition unit 360. The image generation unit 380 uses the high-frequency component representing the high-frequency component cluster stored in the model storage unit 350 in association with the information for specifying the high-frequency component cluster, to generate an image of the object with higher image quality. May be converted to For example, the image generation unit 380 may add the high-frequency component selected for each object with a weight corresponding to the distance from the center of each object to the processing target position on the face to the object image. Note that the representative high-frequency component may be generated by closed-loop learning. As described above, the parameter value calculation unit 260 can select desired learning data for each object from learning data generated by learning for each object. For this reason, since the image generation unit 380 can use the desired learning data selected for each object, the image of the object may be improved in image quality with higher accuracy. In the above example, the information specifying the high frequency component cluster is output from the output unit 207, but the information specifying the low frequency component cluster may be output from the output unit 207. In this case, the model storage unit 350 stores high-frequency clusters in association with information specifying the low-frequency component clusters. Then, the image generation unit 380 associates the high-frequency component representing the high-frequency component cluster stored in the model storage unit 350 in association with the information specifying the low-frequency component cluster output from the output unit 207 into the image of the object. By adding to, the image quality of the object can be improved.

  As described above, according to the image processing device 120 and the image processing device 170, it is possible to reconstruct an image of a feature region based on principal component analysis (PCA). The image reconstruction method related to the image processing device 120 and the image processing device 170 and the learning method for the image reconstruction include learning and image reconstruction by principal component analysis (PCA), as well as locally stored projection ( locality prediction projection (LPP), linear discriminant analysis (LDA), independent component analysis (ICA), multi-dimensional scaling vector support, MDS network, neural network support , Hidden Markov model, Bayesian inference, maximum posterior probability estimation, iterative backprojection, Wavelet transform, local linear embedding locally linear embedding: LLE), Markov random field (Markov random field: MRF) method can be used such as.

  In the above description, the function and operation of each component of the image processing system 10 has been described by taking the case of using a two-dimensional model as an example. However, the image processing system 10 uses a three-dimensional model. May be. That is, the model storage unit 270 and the model storage unit 350 may store a three-dimensional model. In addition, when using a three-dimensional model, it can implement | achieve by adding z component which shows depth to the above-mentioned vector A. FIG. For example, when the vector A is (r1, g1, b1, z1, r2, g2, b2, z1,..., Rm, gm, bm, zm), a three-dimensional model can be realized.

  Note that the three-dimensional models stored in the model storage unit 270 and the model storage unit 350 can be generated using a three-dimensional image generated from a plurality of sample images obtained by imaging an object from different directions. . For example, a three-dimensional model can be generated for each three-dimensional image of a plurality of objects by the same method as that for generating the above-described two-dimensional model. Then, the parameter value calculation unit 260 identifies a feature region including images in which the same object is captured from different directions, among the feature regions in each of the plurality of captured images, and the object value included in the identified feature region. The value of the characteristic parameter is calculated by adapting the 3D image of the object based on the image to the 3D model. Note that the parameter value calculation unit 260 can generate a three-dimensional image of the object based on the parallax information in the image of the same object captured from different directions. Further, based on the parallax information, the direction in which the image of the object included in each feature region is captured is specified. The output unit 207 may output the shooting direction in association with the image of the region other than the feature region and the value of the feature parameter.

  Then, the image generation unit 380 generates a three-dimensional image of the object included in the image of the feature region including the image in which the same object is captured from different directions from the three-dimensional model based on the value of the feature parameter. Based on the obtained three-dimensional image, a two-dimensional image of the object included in the feature region image is generated. The feature area information acquisition unit 360 acquires the shooting direction output from the output unit 207 and acquired by the image acquisition unit 301, and supplies the acquired shooting direction to the image generation unit 380. The image generation unit 380 can generate a two-dimensional image of the object by projecting the image on the two-dimensional space based on the shooting direction and the three-dimensional image. Then, the output unit 340 outputs an image in which the two-dimensional image generated by the image generation unit 380 and the image of the region other than the feature region acquired by the image acquisition unit 301 are combined. The shooting direction is an example of direction information used when generating a two-dimensional image from a three-dimensional image, and the direction information may be a projection angle at which the three-dimensional data is projected onto a two-dimensional space.

  In addition, when the difference between the image of the object included in the captured image and the average image (for example, the average face image) is larger than a predetermined value, the compression unit 230 compresses the feature region image. You can do it. Thereby, when the image of the object included in the feature region is greatly deviated from the average image, it is possible to prevent the reconstruction accuracy from being significantly lowered.

  FIG. 11 shows an example of an image processing system 20 according to another embodiment. The configuration of the image processing system 20 in the present embodiment is the image described with reference to FIG. 1 except that the imaging devices 100a-d have image processing units 804a-d (hereinafter collectively referred to as image processing units 804). The configuration of the processing system 10 is the same.

  The image processing unit 804 has components other than the image acquisition unit 250 among the components included in the image processing apparatus 120. The functions and operations of the constituent elements included in the image processing unit 804 are replaced with that the constituent elements included in the image processing device 120 process the captured moving image obtained by the decompression processing by the compressed moving image decompression unit 202. The functions and operations of the components included in the image processing apparatus 120 may be substantially the same except that the captured moving image captured by the imaging unit 102 is processed. Also in the image processing system 20 having such a configuration, substantially the same effect as that described in relation to the image processing system 10 in FIGS. 1 to 10 can be obtained.

  Note that the image processing unit 804 acquires a captured moving image including a plurality of captured images represented in the RAW format from the imaging unit 102, and a plurality of captured images represented in the RAW format included in the acquired captured moving image (for example, , An image of an area other than the feature area) may be compressed in the RAW format. Note that the image processing unit 804 may detect one or more feature regions from a plurality of captured images expressed in the RAW format. The image processing unit 804 may compress a captured moving image including a plurality of compressed captured images in the RAW format. Note that the image processing unit 804 can compress the captured moving image by the compression method described as the operation of the image processing apparatus 120 in relation to FIGS. Note that the image processing apparatus 170 can acquire a plurality of captured images (for example, images of regions other than the feature region) expressed in the RAW format by expanding the moving image acquired from the image processing unit 804. The image processing apparatus 170 enlarges each of the plurality of captured images expressed in the RAW format acquired by decompression for each region, and performs synchronization processing for each region. At this time, the image processing apparatus 170 may perform synchronization processing with higher accuracy in the feature region than in the region other than the feature region.

  FIG. 12 shows an example of the hardware configuration of the image processing device 120 and the image processing device 170. The image processing device 120 and the image processing device 170 include a CPU peripheral unit, an input / output unit, and a legacy input / output unit. The CPU peripheral section includes a CPU 1505, a RAM 1520, a graphic controller 1575, and a display device 1580 that are connected to each other by a host controller 1582. The input / output unit includes a communication interface 1530, a hard disk drive 1540, and a CD-ROM drive 1560 that are connected to the host controller 1582 by the input / output controller 1584. The legacy input / output unit includes a ROM 1510, a flexible disk drive 1550, and an input / output chip 1570 connected to the input / output controller 1584.

  The host controller 1582 connects the RAM 1520, the CPU 1505 that accesses the RAM 1520 at a higher transfer rate, and the graphic controller 1575. The CPU 1505 operates according to the contents of programs stored in the ROM 1510 and the RAM 1520 to control each unit. The graphic controller 1575 acquires image data generated by the CPU 1505 or the like on a frame buffer provided in the RAM 1520 and displays the image data on the display device 1580. Alternatively, the graphic controller 1575 may include a frame buffer that stores image data generated by the CPU 1505 or the like.

  The input / output controller 1584 connects the host controller 1582 to the hard disk drive 1540, the communication interface 1530, and the CD-ROM drive 1560, which are relatively high-speed input / output devices. The hard disk drive 1540 stores programs and data used by the CPU 1505. The communication interface 1530 is connected to the network communication device 1598 to transmit / receive programs or data. The CD-ROM drive 1560 reads a program or data from the CD-ROM 1595 and provides it to the hard disk drive 1540 and the communication interface 1530 via the RAM 1520.

  The input / output controller 1584 is connected to the ROM 1510, the flexible disk drive 1550, and the relatively low-speed input / output device of the input / output chip 1570. The ROM 1510 stores a boot program that is executed when the image processing device 120 and the image processing device 170 are started, or a program that depends on the hardware of the image processing device 120 and the image processing device 170. The flexible disk drive 1550 reads a program or data from the flexible disk 1590 and provides it to the hard disk drive 1540 and the communication interface 1530 via the RAM 1520. The input / output chip 1570 connects various input / output devices via the flexible disk drive 1550 or a parallel port, serial port, keyboard port, mouse port, and the like.

  A program executed by the CPU 1505 is stored in a recording medium such as the flexible disk 1590, the CD-ROM 1595, or an IC card and provided by the user. The program stored in the recording medium may be compressed or uncompressed. The program is installed in the hard disk drive 1540 from the recording medium, read into the RAM 1520, and executed by the CPU 1505. The program executed by the CPU 1505 causes the image processing apparatus 120 to function as each component included in the image processing apparatus 120 described with reference to FIGS. 1 to 11, and the image processing apparatus 170 is changed to FIGS. It is made to function as each component which image processing apparatus 170 which was explained related has.

  The program shown above may be stored in an external storage medium. As the storage medium, in addition to the flexible disk 1590 and the CD-ROM 1595, an optical recording medium such as a DVD or PD, a magneto-optical recording medium such as an MD, a tape medium, a semiconductor memory such as an IC card, or the like can be used. Further, a storage device such as a hard disk or a RAM provided in a server system connected to a dedicated communication network or the Internet is used as a recording medium, and is provided to the image processing device 120 and the image processing device 170 as a program via the network. Also good. As described above, the computer controlled by the program functions as the image processing device 120 and the image processing device 170.

  Although the present invention has been described using the embodiment, the technical scope of the present invention is not limited to the scope described in the embodiment. It will be apparent to those skilled in the art that various modifications or improvements can be added to the above-described embodiment. It is apparent from the scope of the claims that the embodiments added with such changes or improvements can be included in the technical scope of the present invention.

1 is a diagram illustrating an example of an image processing system 10 according to an embodiment. 2 is a diagram illustrating an example of a block configuration of an image processing apparatus 120. FIG. 3 is a diagram illustrating an example of a block configuration of a compression unit 230. FIG. 2 is a diagram illustrating an example of a block configuration of an image processing apparatus 170. FIG. It is a figure which shows an example of the other block structure of the compression part 230. FIG. It is a figure which shows an example of the feature point in a human face. It is a figure which shows typically an example of the change of the face shape at the time of changing the weighting coefficient b. It is a figure which shows an example of the image obtained by converting a sample image into an average face shape. It is a figure which shows typically an example of the change of the pixel value at the time of changing the weighting coefficient q. It is a figure which shows an example of the model which the model storage part 270 and the model storage part 350 have stored in the table format. It is a figure which shows an example of the image processing system 20 which concerns on other embodiment. 2 is a diagram illustrating an example of a hardware configuration of an image processing device 120 and an image processing device 170. FIG.

DESCRIPTION OF SYMBOLS 10 Image processing system 20 Image processing system 100 Image pick-up apparatus 102 Image pick-up part 104 Image pick-up image compression part 110 Communication network 120 Image processing apparatus 130 Person 140 Mobile body 150 Monitoring object space 160 Space 170 Image processing apparatus 175 Image DB
180 Display Device 201 Compressed Movie Acquisition Unit 202 Compressed Movie Expansion Unit 203 Feature Area Detection Unit 206 Association Processing Unit 207 Output Unit 210 Compression Control Unit 230 Compression Unit 232 Image Dividing Unit 234 Fixed Value Unit 236 Compression Processing Unit 240 Image Quality Conversion Unit 241 Image quality conversion unit 250 Image acquisition unit 260 Parameter value calculation unit 270 Model storage unit 280 Parameter quantization unit 301 Image acquisition unit 302 Association analysis unit 310 Decompression control unit 320 Decompression unit 322 Decoder 330 Synthesis unit 332 Enlargement unit 340 Output unit 350 Model storage unit 360 Feature region information acquisition unit 380 Image generation unit 510 Image quality conversion unit 520 Difference processing unit 522 Inter-layer difference processing unit 530 Encoding unit 532 Encoder 804 Image processing unit 1505 CPU
1510 ROM
1520 RAM
1530 Communication interface 1540 Hard disk drive 1550 Flexible disk drive 1560 CD-ROM drive 1570 Input / output chip 1575 Graphic controller 1580 Display device 1582 Host controller 1584 Input / output controller 1590 Flexible disk 1595 CD-ROM
1598 Network communication device

Claims (18)

A model storage unit for storing a model in which an object is represented by a feature parameter;
A feature region detector for detecting a feature region from the captured image;
A feature in the model that represents an object included in the image of the feature region by adapting an image of the object included in the image of the feature region in the captured image to the model stored in the model storage unit. A parameter value calculation unit for calculating a parameter value;
An image processing system comprising: a feature parameter value calculated by the parameter value calculation unit; and an output unit that outputs an image of a region other than the feature region. A compression unit that compresses the image of the region other than the feature region by reducing the image quality of the image of the region other than the feature region;
The image processing system according to claim 1, wherein the output unit outputs the value of the characteristic parameter calculated by the parameter value calculation unit and an image of an area other than the characteristic area whose image quality has been reduced by the compression unit. The image processing system according to claim 2, wherein the model storage unit stores the model in which an object is expressed by a statistical feature parameter. The model storage unit stores the model in which an object is represented by principal components based on principal component analysis,
The parameter value calculation unit calculates a weight of a principal component in the model,
The image processing system according to claim 3, wherein the output unit outputs an image of a region other than the weight of the principal component calculated by the parameter value calculation unit and the feature region. The image processing system according to claim 4, wherein the model storage unit stores the model in which a shape of an object is expressed by a principal component based on principal component analysis. The image processing system according to claim 4, wherein the model storage unit stores the model in which a color of an object is expressed by a principal component based on principal component analysis. The model storage unit stores the models of different types of objects in association with the types,
The parameter value calculation unit stores the model image stored in the model storage unit in association with the type of the object imaged in the feature region, in the image of the feature region in the captured image. The image processing system according to claim 3, wherein the value of the feature parameter is calculated by adapting to the method. The output unit outputs the value of the feature parameter calculated by the parameter value calculation unit, the type of object imaged in the feature region, and an image of a region other than the feature region whose image quality has been reduced by the compression unit. The image processing system according to claim 7. The model storage unit stores the model of the object viewed from different directions in association with the direction,
The parameter value calculation unit stores an image of an object included in the image of the feature region in the captured image in association with a direction in which the object imaged in the feature region is captured, and stores the model storage unit. The image processing system according to claim 3, wherein the characteristic parameter value is calculated by adapting to the model. The output unit includes a value of a feature parameter calculated by the parameter value calculation unit, a direction in which an object imaged in the feature region is imaged, and a region other than the feature region whose image quality has been reduced by the compression unit. The image processing system according to claim 9 which outputs an image. The model storage unit stores the model of an object illuminated under different illumination conditions in association with the illumination condition,
The parameter value calculation unit stores an image of an object included in the image of the feature area in the captured image in association with an illumination condition in which the object imaged in the feature area is illuminated. The image processing system according to claim 3, wherein the characteristic parameter value is calculated by adapting to the model. The output unit includes a feature parameter value calculated by the parameter value calculation unit, an illumination condition in which an object imaged in the feature region is illuminated, and a region other than the feature region in which image quality is reduced by the compression unit The image processing system according to claim 11, wherein the image is output. The image processing system according to claim 3, wherein the compression unit compresses an image in a region other than the feature region by reducing the resolution of the image in a region other than the feature region. A parameter quantization unit that selects a feature parameter from which the value of the feature parameter is to be output from the output unit;
The image processing system according to claim 3, wherein the output unit outputs a value of a feature parameter selected by the parameter quantization unit and an image of a region other than the feature region whose image quality has been reduced by the compression unit. An image acquisition unit that acquires an image of a region other than the feature region in which the value of the feature parameter and the image quality output from the output unit are reduced;
An image generation unit that generates an image of an object included in the image of the feature region from the model based on a value of a feature parameter;
The image processing system according to claim 3, further comprising: an image output unit that outputs an image obtained by combining the image of the object generated from the model and an image of an area other than the feature area acquired by the image acquisition unit. . The parameter value calculation unit captures an image of the object included in the image of the feature region in the captured image by being applied to the model stored in the model storage unit, thereby capturing the image of the feature region. Calculating the value of the feature parameter in the model, representing the shape of the object
The compression unit compresses the captured image by reducing the image quality of the feature region and the region other than the feature region in the captured image,
The output unit outputs the value of the characteristic parameter calculated by the parameter value calculation unit, and the captured image whose image quality is reduced by the compression unit,
The image acquisition unit acquires the value of the characteristic parameter output from the output unit and the captured image with the image quality reduced,
The image generation unit generates the shape of the object captured in the image of the feature region from the model based on the value of the feature parameter, and the shape of the generated object and the captured image acquired by the image acquisition unit The image processing system according to claim 15, wherein an image of an object captured in the feature region image is generated using a pixel value of the feature region image. A feature region detection stage for detecting a feature region from the captured image;
A model storage stage for storing a model in which an object is represented by feature parameters;
The model representing an object captured in the image of the feature region by adapting an image of the object included in the image of the feature region in the captured image to the model stored in the model storage step. A parameter value calculation stage for calculating the value of the characteristic parameter in
An image processing method comprising: a feature parameter value calculated in the parameter value calculation step and an output step of outputting an image of a region other than the feature region. A program for an image processing system, comprising:
A feature region detection unit for detecting a feature region from the captured image;
A model storage unit for storing a model in which an object is represented by a feature parameter;
The model representing an object imaged in the image of the feature region by adapting an image of the object included in the image of the feature region in the captured image to the model stored in the model storage unit A parameter value calculation unit for calculating the value of the characteristic parameter in
A program that functions as an output unit that outputs a value of a characteristic parameter calculated by the parameter value calculation unit and an image of an area other than the characteristic area.

Images