Abstract: An image processing device according to the present invention includes: a patch generation unit which generates an input patch used for comparison on the basis of an input image; a modification parameter estimation unit which estimates a parameter used in blurred modification on the basis of the input image; a blurred image generation unit which generates a blurred image on the basis of a learning image by using the parameter; a patch pair generation unit which generates a patch pair used to compose a restoration image on the basis of the blurred image and the learning image; a selection unit which selects a patch pair used to compose the restoration image on the basis of the input patch; and a composition unit which composes the restoration image on the basis of the patch pair selected by the selection unit.

Abstract: The present invention provides an image processing device whereby the probability of outputting a restored image which accurately corresponds to an original image which is included in a low-quality input image is improved. This image processing device comprises: an image group generating means for generating, from the input image, using a dictionary which stores a plurality of patch pairs wherein a degradation patch which is a patch of a degraded image wherein a prescribed image is degraded is associated with a restoration patch which is a patch of this prescribed image, a plurality of restored image candidates including a plurality of different instances of content which have a possibility of being the original content of the input image; and an image selection presentation means for clustering the generated plurality of restored image candidates, and selecting and outputting an image candidate on the basis of the result of this clustering.

Abstract: Provided is an image processing device that can bring about the sufficient resemblance between an original image and a restored image obtained corresponding to a low resolution input image. The image processing device includes a means that uses a dictionary for storing data associating deteriorated patches which are from a deteriorated image formed by deteriorating a prescribed image, and restoration patches which are from the prescribed image, and calculates, as a degree-of-similarity between plural input patches generated by dividing an input image and the deteriorated patches, a weighted degree-of-similarity between weighted deteriorated patches and weighted input patches, in which forms of the deteriorated patches and the input patches are reconfigured using a patch weight which is continuous weighting; a means that selects, on the basis of the weighted degree-of-similarity, a restoration patch for each input patch; and a means that combines the restoration patches to generate a restored image.

Abstract: There is provided an image dynamic range compression system which can compress a dynamic range for which the visibility of a low-frequency image is ensured while preserving a high-frequency image. An image converting unit converts an input image into a compressed image having a narrower dynamic range than the input image. A high-frequency image extracting unit extracts a high-frequency image from the input image. An image synthesizing unit synthesizes a compressed image and the high-frequency image. Further, by adaptively changing synthesizing method of ensuring the visibility of images to synthesize, the image synthesizing unit synthesizes these images.

Abstract: Provided is a motion vector estimation device capable of estimating the motion vector with less computation. A motion vector estimation device for estimating, by means of repetitive calculations, the motion vector of each of a plurality of pixel groups which is contained in an input image and which each contains one or more pixels, the motion vector estimation device being provided with a means for making repetitive calculations with regard to the pixel groups that do not have a high frequency component from among the plurality of pixel groups contained in the input image after making repetitive calculations with regard to the pixel groups that have a high frequency component from among the plurality of pixel groups contained in the input image.

Abstract: Error-function determining means makes use of a blur function serving as a function representing the degree of blurring of a blurred input image and a geometrical-deformation matrix serving as a matrix for restricting an operation to lay out a 2-dimensional code shown by the input image on cells each serving as a configuration unit of the 2-dimensional code in order to determine an error function serving as a function satisfying a relation that, the smaller the error between the input image and an image obtained by adding geometrical deformations and blurs to a recovered image obtained from a recovering process, the more the approach to a result determined in advance. Pixel-value restricting means determines pixel-value restriction information prescribing a restriction condition for a restriction imposed on a predetermined pixel value obtained from discretization of pixel values of the 2-dimensional code to be recovered.

Abstract: For each pixel of an input image, a direction toward which variation in pixel value between the relevant pixel and its peripheral pixels is largest and an amount of the variation in pixel value are computed, where a direction unit vector that indicates the direction toward which the variation is largest and a variation vector that represents the amount of the largest variation in the pixel value are computed. For each pixel, a regularization strength is computed utilizing the direction unit vector, the variation vector, and a regularization strength computing formula so that the larger the amount of the variation in pixel value toward a direction indicated by the direction unit vector, the less the regularization strength. An optimization function is determined based on the input image and the regularization strength for each pixel thereof and assigns a value for each pixel, by which the determined optimization function has the minimum value, to the corresponding pixel of a restored image to be generated.

Abstract: An automatic projection image correction system (1) includes: a projection image frame detection unit (101) that detects straight lines forming a frame of a screen-projected image which is projected onto a screen by a projector from a captured image obtained by capturing the screen-projected image using a camera; and a projector plane parameter estimation unit (102) that sets coordinate axes to a projector image, which is an original image of the screen-projected image, and the captured image and estimates at least one of a distance between the screen and the projector, an inclination of the screen with reference to an optical axis of the projector, and a zoom ratio of the projector, using an inclination of the optical axes of the projector and the camera, a distance between the optical centers of the projector and the camera, an intrinsic parameter of the camera, coordinate data of the detected straight lines at coordinates of the captured image, and coordinate data of straight lines corresponding to the fra

Abstract: The present invention generates a highly-accurate restored image while reducing computation costs by obtaining a similar effect to when the adjacent similarity of restoration patches is taken into consideration.

Abstract: Provided is a motion vector estimation device capable of estimating the motion vector with less computation. A motion vector estimation device for estimating, by means of repetitive calculations, the motion vector of each of a plurality of pixel groups which is contained in an input image and which each contains one or more pixels, the motion vector estimation device being provided with a means for making repetitive calculations with regard to the pixel groups that do not have a high frequency component from among the plurality of pixel groups contained in the input image after making repetitive calculations with regard to the pixel groups that have a high frequency component from among the plurality of pixel groups contained in the input image.

Abstract: A dictionary creation device including a blurred image generation unit which outputs a blurred image generated by performing a blurring process to a learning image together with a blur parameter indicating a blurring state of the blurred image, a patch pair generation unit which generates a restoration patch and a blurred patch as a patch pair that is composed of the patches located at the corresponding positions of the learning image and the blurred image, and a registration unit which associates the patch pair with a blur parameter corresponding to the blurred patch in the patch pair and registers them in a dictionary.

Abstract: A geometric transformation parameter computing unit computes a geometric transformation parameter which represents the geometric transformation of an image from the previous frame to the current frame, and calculates the level of reliability in that geometric transformation parameter. A super-resolution image prediction unit generates a prediction of a super-resolution image in the current frame by transforming a super-resolution image in the previous frame on the basis of the geometric transformation parameter. A super-resolution image generation unit calculates a low-resolution image in the current frame from the prediction result of the super-resolution image in the current frame by a simulation, calculates a difference between a low-resolution image, which is an input image for the current frame, and the simulation result, and calculates a weighted mean between a result of upsampling the difference and the prediction result of the super-resolution image in the current frame.

Abstract: Provided is an image processing system that can uniquely identify a motion vector for each pixel of an interpolation frame and accurately identify a motion vector. A motion vector candidate selection unit (301) sequentially selects predetermined motion vector candidates in an interpolation frame. A differential image generation unit (302), a binary image generation unit (303) and an area image generation unit (304) generate a differential image, a binary image and an area image, respectively, based on a selected motion vector candidate. For each pixel of the interpolation frame, a motion vector determination unit (305) identifies a motion vector candidate with which a pixel value of a pixel in the area image corresponding to the pixel is the largest as a motion vector. A pixel value determination unit (306) determines each pixel value of the interpolation frame by using the motion vector identified for each pixel, thereby generating an interpolation frame.

Abstract: [PROBLEMS] In a system for converting a low resolution image to a high resolution image, it is difficult to carry out processing to generate a high resolution image for each frame in real time using temporally continuous frame images. [MEANS FOR SOLVING THE PROBLEMS] An image processing system includes a conversion parameter calculation unit 101, an initial image generation unit 102, a high resolution image generation unit 103, and an input image and conversion parameter storage unit 110, and converts the high resolution image generated at the previous frame in accordance with a current frame based on the conversion parameter for the current frame with respect to the previous frame so as to perform high resolution image generation processing of the current frame with the converted image as an initial value.

Abstract: There is provided an image processing system which can remove jaggies not only from binary images but also from gray scale images, color images and images upscaled by an unknown scaler. When the center pixel of the local area clipped from a processing target image is decided to be the jaggy removal target pixel, a base filter computing means 84 applies a plurality of types of predetermined base filters to the local area. A weight calculating means 85 calculates weights with respect to the plurality of types of base filters based on the feature amount calculated by a principal component analyzing means 82. Based on a filtering result of each base filter and a weight with respect to each base filter, a filtering result weight computing means 86 updates a pixel value of the center pixel.

Abstract: Error-function determining means makes use of a blur function serving as a function representing the degree of blurring of a blurred input image and a geometrical-deformation matrix serving as a matrix for restricting an operation to lay out a 2-dimensional code shown by the input image on cells each serving as a configuration unit of the 2-dimensional code in order to determine an error function serving as a function satisfying a relation that, the smaller the error between the input image and an image obtained by adding geometrical deformations and blurs to a recovered image obtained from a recovering process, the more the approach to a result determined in advance. Pixel-value restricting means determines pixel-value restriction information prescribing a restriction condition for a restriction imposed on a predetermined pixel value obtained from discretization of pixel values of the 2-dimensional code to be recovered.

Abstract: Even when an enlarged image is deteriorated due to a block noise or color mixture, the contour in the enlarged image is sharpened and a quality of the image is improved. A contour extracting unit 103 determines a pixel indicating the contour in an interpolation image, using brightness gradient strength and a brightness gradient direction.

Abstract: There is provided an image dynamic range compression system which can compress a dynamic range for which the visibility of a low-frequency image is ensured while preserving a high-frequency image. An image converting unit converts an input image into a compressed image having a narrower dynamic range than the input image. A high-frequency image extracting unit extracts a high-frequency image from the input image. An image synthesizing unit synthesizes a compressed image and the high-frequency image. Further, by adaptively changing synthesizing method of ensuring the visibility of images to synthesize, the image synthesizing unit synthesizes these images.

Abstract: In an image processing apparatus, a conditional probability calculating element calculates the probability of existence of an attribute as to each region in an input degraded image, for each attribute. Optimization function determining element uses a conditional probability calculated by the conditional probability calculating element with regard to each region in the degraded image for each attribute, a-priori knowledge defined as a conditional probability of appearance of an attribute in a region of a candidate of a restored image when there is the attribute in the region, and the input degraded image to determine an optimization function using the candidate of the restored image as a variable and yielding a smaller value as the candidate of the restored image is closer to a true restored image. Restored image determining element adopts the candidate of the restored image with which the optimization function yields a minimum value as the restored image.

Abstract: A geometric transformation parameter computing unit computes a geometric transformation parameter which represents the geometric transformation of an image from the previous frame to the current frame, and calculates the level of reliability in that geometric transformation parameter. A super-resolution image prediction unit generates a prediction of a super-resolution image in the current frame by transforming a super-resolution image in the previous frame on the basis of the geometric transformation parameter. A super-resolution image generation unit calculates a low-resolution image in the current frame from the prediction result of the super-resolution image in the current frame by a simulation, calculates a difference between a low-resolution image, which is an input image for the current frame, and the simulation result, and calculates a weighted mean between a result of upsampling the difference and the prediction result of the super-resolution image in the current frame.