Abstract: To enable obtainment of a composite image including a movement trajectory of a moving body, even from a moving image including a background. A sampler samples frames from moving image data. A first mask frame calculator calculates mask frames including mask regions, which correspond to positions of the moving body on the frames. A second mask frame calculator accumulates and binarizes the mask frames to calculate a reference mask frame. A region masker masks the frames by using the mask frames and the reference mask frame to cut out moving body regions, which correspond to a moving body, from the frames. A composer overwrites moving body regions on one of the frames in chronological order to obtain a composite image, which includes the movement trajectory of the moving body.

Abstract: To acquire a high-resolution frame from a plurality of frames sampled from a video image, it is necessary to obtain a high-resolution frame with reduced picture quality degradation regardless of motion of a subject included in the frame. Because of this, between a plurality of contiguous frames FrN and FrN+1, there is estimated a correspondent relationship. Based on the correspondent relationship, the frames FrN+1 and FrN are interposed to obtain first and second interpolated frames FrH1 and FrH2. Based on the correspondent relationship, the coordinates of the frame FrN+1 are transformed, and from a correlation value with the frame FrN, there is obtained a weighting coefficient ? (x?, y?) that makes the weight of the first interpolated frame FrH1 greater as a correlation becomes greater. With the weighting coefficient, the first and second interpolated frames are weighted and added to acquire a synthesized frame FrG.

Abstract: Disclosed is a method for accurately and efficiently detecting a modality of input data, including the steps of projecting the input data into a plurality of projection data using each of a plurality of transformation matrix groups U1·(?12U2T), generating a plurality of inverse projection data by performing inverse projection of the transformation matrix groups on the plurality of generated projection data, calculating a correlation between the input data and the generated inverse projection data with respect to each transformation matrix group U1·(?12U2T), and identifying a modality represented by a transformation matrix group having a highest calculated correlation as the modality of the input data.

Abstract: When NN pairs of sample data are inputted, pre-processing and area division processing are performed by a pre-processing means and area division means respectively. Thereafter, when number of classes NP is inputted to projection matrix generation means based on the NN pairs of sample data, clustering in which NN pairs of sample data are classified into NP classes. Then, representative sample data representing each of classes (NP classes) are selected from NN pairs of sample data as the learning data. Then, based on the nature of Formula (2), calculation means generates an inverse first projection matrix (ML)?1 using a plurality of low resolution sample images (pre-transformation sample data) of a plurality of pairs of representative sample data and a second projection matrix MH using a plurality of high resolution sample images (post-transformation sample data). Thereafter, a transformation matrix {MH·(ML)1} is generated by transformation matrix generation means.

Abstract: A bright luminance value, a dark luminance value, and an average luminance value relating to a face image portion included in an image represented by fed image data are calculated in calculating circuits. Further, a target bright luminance value and a target dark luminance value are calculated in a calculating circuit on the basis of a target average luminance value and a dynamic range that are inputted from an input device and the calculated bright luminance value, dark luminance value, and average luminance value. Interpolation processing based on a correspondence between the calculated bright luminance value, dark luminance value, and average luminance value relating to the face image portion and the target bright luminance value, the target dark luminance value, and the target average luminance value respectively corresponding thereto is performed in a corrected value calculating circuit, to create a look-up table.

Abstract: In an aspect of an image processing technique according to the present invention, a high-frequency component is extracted from a low-resolution image z which is a restoration target, and a high-frequency image z? is generated (#20). Interpolation filter coefficients A?i, B?i and ?i and a representative high-frequency image z?i are set (#22), a Gaussian mixture model represented by x=?{((A?i×z?)+B?i)×wi((z?i?z?), ?i)} is assumed and a filtering process is applied to the high-frequency image z? (#24). On the other hand, an enlargement process is applied to a low-frequency component and a medium-frequency component of the low-resolution image z (#26), a processing result of the filtering process and a processing result of the enlargement process are added (#28), and a high-resolution image is restored.

Abstract: In an image processing method for carrying out an image quality correction for correcting an image quality on each of a plurality of images in an image group to obtain corrected images, an image processing method includes the steps of carrying out a temporary image quality correction for correcting an image quality on each images in the image group to obtain a temporarily corrected image. A characteristic value representing an image quality for each temporarily corrected image is calculated and a target characteristic value is calculated so that dispersion of each characteristic value to the target characteristic value is minimized on the basis of the characteristic value of each temporarily corrected image. A correction value of each of the plurality of images in the image group which corrects the characteristic value thereof to the target characteristic value is obtained, and the image is corrected with the correction value.

Abstract: On the basis of image data representing the area of a face image contained in an image represented by applied image data, a color correction value calculating circuit calculates a color correction value and a color image probability calculating circuit calculates the probability that the area of the face image is a color image. A color correction value adjusting circuit then adjusts the color correction value based upon the probability that the area of the face image is a color image. The adjusted color correction value is used in correction processing in an image correcting circuit.

Abstract: On the basis of image data representing the area of a face image contained in an image represented by applied image data, a color correction value calculating circuit calculates a color correction value and a color image probability calculating circuit calculates the probability that the area of the face image is a color image. A color correction value adjusting circuit then adjusts the color correction value based upon the probability that the area of the face image is a color image. The adjusted color correction value is used in correction processing in an image correcting circuit.

Abstract: A system is provided to compress an image of a subject captured in a plurality of directions, at high compression, and the image processing apparatus includes: a model storage section that stores a reference model that is a three-dimensional model representing an object; a model generating section that generates, based on a plurality of captured images of an object, an object model that is a three-dimensional model that matches the object captured in the plurality of captured images; and an output section that outputs a position and a direction of the object captured in each of the plurality of captured images, in association with difference information between the reference model and the object model.

Abstract: The present invention determines the adopting ratio (weight coefficient) between the high image quality processing using the tensor projection method and the high image quality processing using another method according to the degree of deviation of the input condition of the input image, and combines these processes as appropriate. This allows a satisfactory reconstruction image to be acquired even in a case of deviation from the input condition, and avoids deterioration of the high quality image due to deterioration of the reconstruction image by the projective operation.

Abstract: The image processing apparatus and method, and the program and the recording medium according to the present invention can make the coefficient vector into high precision by noise elimination or correction utilizing the mutual correlation of the divided image areas in the intermediate eigenspace, and allows relaxation of the input condition and robustness. The high correlation in the divided image areas in the intermediate eigenspace can reduce the divided image areas to be processed, and actualize reduction in processing load and enhancement of the processing speed.

Abstract: An image processing system providing a monitor image in which an image of a characteristic region is easily viewed. The image processing system includes an image obtaining section that obtains a moving image; a characteristic region information obtaining section that obtains information indicating positions of characteristic regions in a plurality of moving image constituting images included in the moving image; and an image generating section that generates display images having substantially the same size as each other, by reducing or enlarging each of images of the characteristic regions included in the plurality of moving image constituting images, based on the positions indicated by the information obtained by the characteristic region information obtaining section.

Abstract: An image processing system capable of reducing an operation time required to generate a high image quality image, includes an image obtaining section that obtains an input image having been captured; a characteristic region information obtaining section that obtains information indicating a characteristic region in the input image; a model storage section that stores a model representing an object by a character parameter; an image generating section that converts an image of an object included in the characteristic region in the input image into a high image quality image having an image quality higher than an image quality of the input image, by adapting the image of the object included in the characteristic region in the input image to the model; and an output section that outputs an image including the high image quality image and an image other than the characteristic region.

Abstract: An image processing system can quickly calculate a motion vector. The image processing system includes: a characteristic region detecting section that detects a characteristic region from each of a plurality of moving image constituting images included in a moving image; a characteristic region identifying section that identifies characteristic regions having high correlation in the plurality of moving image constituting images; a position difference calculating section that calculates a difference in position between the characteristic regions having high correlation identified by the characteristic region identifying section; a search region determining section that determines a narrower motion vector search region when a magnitude of the difference in position is smaller; and a compression section that compresses the moving image using a motion vector calculated by searching the motion vector search region determined by the search region determining section.

Abstract: The present invention provides a video in which a subject image looks clear while reducing the amount of transmitted data. An image capturing apparatus includes: an image capturing section that successively captures a plurality of images under a plurality of image capturing conditions different from each other; an image generating section that generates an output image by overlapping the plurality of captured images captured under the plurality of image capturing conditions; and an output section that outputs the output image generated by the image generating section. The image capturing section may successively capture the plurality of images by means of exposure in different exposure time lengths from each other. The image capturing section may successively capture the plurality of images also by means of exposure in different aperture openings from each other.

Abstract: A trimmed image is obtained by trimming for cutting out from a photographic image obtained by taking a photograph of an object including a reference part, and one or more other part which has a common centerline with the reference part and is arranged in a direction in which the centerline of the reference part extends, an image of an area including a part or the whole of the reference part and said other part along a trimming centerline. The centerline of the reference part is obtained and the degree and the direction of inclination of said other part are detected from the photographic image. The trimming centerline is inclined by a larger angle away from the centerline toward the direction of the inclination of said other part as the degree of inclination increases.

Abstract: Provided is an image processing apparatus including a parameter storage that stores, in association with a plurality of attributes, a plurality of image processing parameters for respectively rendering subject images of different attributes from each other, into high image quality; and an image generating section that generates a high quality image of a subject image included in an input image, using the plurality of image processing parameters stored in the parameter storage. The image generating section may select at least one image from high quality images of the subject image included in the input image using respectively different combinations of the plurality of image processing parameters, based on comparison with the input image, and generate the selected image as the high quality image.

Abstract: To enable obtainment of a composite image including a movement trajectory of a moving body, even from a moving image including a background. A sampler samples frames from moving image data. A first mask frame calculator calculates mask frames including mask regions, which correspond to positions of the moving body on the frames. A second mask frame calculator accumulates and binarizes the mask frames to calculate a reference mask frame. A region masker masks the frames by using the mask frames and the reference mask frame to cut out moving body regions, which correspond to a moving body, from the frames. A composer overwrites moving body regions on one of the frames in chronological order to obtain a composite image, which includes the movement trajectory of the moving body.

Abstract: To acquire a high-resolution frame from a plurality of frames sampled from a video image, it is necessary to obtain a high-resolution frame with reduced picture quality degradation regardless of motion of a subject included in the frame. Because of this, between a plurality of contiguous frames FrN and FrN+1, there is estimated a correspondent relationship. Based on the correspondent relationship, the frames FrN+1 and FrN are interposed to obtain first and second interpolated frames FrH1 and FrH2. Based on the correspondent relationship, the coordinates of the frame FrN+1 are transformed, and from a correlation value with the frame FrN, there is obtained a weighting coefficient ?(x?, y?) that makes the weight of the first interpolated frame FrH1 greater as a correlation becomes greater. With the weighting coefficient, the first and second interpolated frames are weighted and added to acquire a synthesized frame FrG.