Image processing apparatus and method, data processing apparatus and method, and program and recording medium
First Claim
Patent Images
1. An image processing apparatus, comprising:
- an information acquisition device for acquiring an eigen projective matrix generated by a projective operation from a learning image group including pairs of first quality images and second quality images different in image quality from each other, and a projective kernel tensor generated from the learning image group and the eigen projective matrix;
a first sub-kernel tensor generation device for generating a first sub-kernel tensor satisfying a condition specified by a first setting from the acquired projective kernel tensor;
a second sub-kernel tensor generation device for generating a second sub-kernel tensor satisfying a condition specified by a second setting from the acquired projective kernel tensor;
a first sub-tensor projection device for projecting an input image as a processing target using a first projective operation utilizing the eigen projective matrix and the first sub-kernel tensor, and calculating a coefficient vector in an intermediate eigenspace;
a second sub-tensor projection device for generating a modified image having a different image quality from that of the input image by projecting the calculated coefficient vector using a second projective operation utilizing the second sub-kernel tensor and the eigen projective matrix;
a high image quality processing device for generating a high quality image having the same size as that of the modified image from the input image;
a learning image coefficient vector acquisition device for acquiring a coefficient vector of the learning image in the intermediate eigenspace;
a weight coefficient determination device for determining a weight coefficient according to a mutual relationship between the coefficient vector of the learning image in the intermediate eigenspace and the coefficient vector of the input image in the intermediate eigenspace calculated by the first sub-tensor projection device; and
a synthesis device for determining an adopting ratio between a process including the first and second projective operations and a process by the high image quality processing device according to the weight coefficient, and combining the modified image and the high quality image.
1 Assignment
0 Petitions
Accused Products
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.
14 Citations
31 Claims
-
1. An image processing apparatus, comprising:
-
an information acquisition device for acquiring an eigen projective matrix generated by a projective operation from a learning image group including pairs of first quality images and second quality images different in image quality from each other, and a projective kernel tensor generated from the learning image group and the eigen projective matrix; a first sub-kernel tensor generation device for generating a first sub-kernel tensor satisfying a condition specified by a first setting from the acquired projective kernel tensor; a second sub-kernel tensor generation device for generating a second sub-kernel tensor satisfying a condition specified by a second setting from the acquired projective kernel tensor; a first sub-tensor projection device for projecting an input image as a processing target using a first projective operation utilizing the eigen projective matrix and the first sub-kernel tensor, and calculating a coefficient vector in an intermediate eigenspace; a second sub-tensor projection device for generating a modified image having a different image quality from that of the input image by projecting the calculated coefficient vector using a second projective operation utilizing the second sub-kernel tensor and the eigen projective matrix; a high image quality processing device for generating a high quality image having the same size as that of the modified image from the input image; a learning image coefficient vector acquisition device for acquiring a coefficient vector of the learning image in the intermediate eigenspace; a weight coefficient determination device for determining a weight coefficient according to a mutual relationship between the coefficient vector of the learning image in the intermediate eigenspace and the coefficient vector of the input image in the intermediate eigenspace calculated by the first sub-tensor projection device; and a synthesis device for determining an adopting ratio between a process including the first and second projective operations and a process by the high image quality processing device according to the weight coefficient, and combining the modified image and the high quality image. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22)
-
-
23. An image processing apparatus, comprising:
-
an information acquisition device for acquiring an eigen projective matrix generated by a projective operation from a learning image group including pairs of first quality images and second quality images different in image quality from each other, a first sub-kernel tensor satisfying a condition specified by a first setting generated using a projective kernel tensor generated from the learning image group and the eigen projective matrix, and a second sub-kernel tensor satisfying a condition specified by a second setting generated using the projective kernel tensor; a first sub-tensor projection device for projecting an input image as a processing target using a first projective operation utilizing the eigen projective matrix and the first sub-kernel tensor, and calculating a coefficient vector in an intermediate eigenspace; a second sub-tensor projection device for generating a modified image having a different image quality from that of the input image by projecting the calculated coefficient vector using a second projective operation utilizing the second sub-kernel tensor and the eigen projective matrix; a high image quality processing device for generating a high quality image having the same size as that of the modified image from the input image; a learning image coefficient vector acquisition device for acquiring a coefficient vector of the learning image in the intermediate eigenspace; a weight coefficient determination device for determining a weight coefficient according to a mutual relationship between the coefficient vector of the learning image in the intermediate eigenspace and the coefficient vector of the input image in the intermediate eigenspace calculated by the first sub-tensor projection device; and a synthesis device for determining an adopting ratio between a process including the first and second projective operations and a process by the high image quality processing device according to the weight coefficient, and combining the modified image and the high quality image.
-
-
24. An image processing apparatus, comprising:
-
an eigen projective matrix generation device for generating an eigen projective matrix generated by a projective operation from a learning image group including pairs of first quality images and second quality images different in image quality from each other; a projective kernel tensor generation device for generating a projective kernel tensor specifying a corresponding relationship between the first quality images and an intermediate eigenspace and a corresponding relationship between the second quality images and the intermediate eigenspace; a first sub-kernel tensor acquisition device for generating a first sub-kernel tensor satisfying a condition specified by a first setting from the generated projective kernel tensor; a second sub-kernel tensor acquisition device for generating a second sub-kernel tensor satisfying a condition specified by a second setting from the generated projective kernel tensor; a first sub-tensor projection device for projecting an input image as a processing target using a first projective operation utilizing the eigen projective matrix and the first sub-kernel tensor, and calculating a coefficient vector in the intermediate eigenspace; a second sub-tensor projection device for generating a modified image having a different image quality from that of the input image by projecting the calculated coefficient vector using a second projective operation utilizing the second sub-kernel tensor and the eigen projective matrix; a high image quality processing device for generating a high quality image having the same size as that of the modified image from the input image; a learning image coefficient vector acquisition device for acquiring a coefficient vector of the learning image in the intermediate eigenspace; a weight coefficient determination device for determining a weight coefficient according to a mutual relationship between the coefficient vector of the learning image in the intermediate eigenspace and the coefficient vector of the input image in the intermediate eigenspace calculated by the first sub-tensor projection device; and a synthesis device for determining an adopting ratio between a process including the first and second projective operations and a process by the high image quality processing device according to the weight coefficient, and combining the modified image and the high quality image. - View Dependent Claims (25)
-
-
26. An image processing method, including:
-
an information acquiring step of acquiring an eigen projective matrix generated by a projective operation from a learning image group including pairs of first quality images and second quality images different in image quality from each other, and a projective kernel tensor generated from the learning image group and the eigen projective matrix; a first sub-kernel tensor generating step of generating a first sub-kernel tensor satisfying a condition specified by a first setting from the acquired projective kernel tensor; a second sub-kernel tensor generating step of generating a second sub-kernel tensor satisfying a condition specified by a second setting from the acquired projective kernel tensor; a first sub-tensor projecting step of projecting an input image as a processing target using a first projective operation utilizing the eigen projective matrix and the first sub-kernel tensor, and calculating a coefficient vector in an intermediate eigenspace; a second sub-tensor projecting step of generating a modified image having a different image quality from that of the input image by projecting the calculated coefficient vector using a second projective operation utilizing the second sub-kernel tensor and the eigen projective matrix; a high image quality processing step of generating a high quality image having the same size as that of the modified image from the input image; a learning image coefficient vector acquiring step of acquiring a coefficient vector of the learning image in the intermediate eigenspace; a weight coefficient determining step of determining a weight coefficient according to a mutual relationship between the coefficient vector of the learning image in the intermediate eigenspace and the coefficient vector of the input image in the intermediate eigenspace calculated by the first sub-tensor projecting step; and a synthesis step of determining an adopting ratio between a process including the first and second projective operations and a process by the high image quality processing step according to the weight coefficient, and combining the modified image and the high quality image.
-
-
27. An image processing method, including:
-
an information acquiring step of acquiring an eigen projective matrix generated by a projective operation from a learning image group including pairs of first quality images and second quality images different in image quality from each other, a first sub-kernel tensor satisfying a condition specified by a first setting generated using a projective kernel tensor generated from the learning image group and the eigen projective matrix, and a second sub-kernel tensor satisfying a condition specified by a second setting generated using the projective kernel tensor; a first sub-tensor projecting step of projecting an input image as a processing target using a first projective operation utilizing the eigen projective matrix and the first sub-kernel tensor, and calculating a coefficient vector in an intermediate eigenspace; a second sub-tensor projecting step of generating a modified image having a different image quality from that of the input image by projecting the calculated coefficient vector using a second projective operation utilizing the second sub-kernel tensor and the eigen projective matrix; a high image quality processing step of generating a high quality image having the same size as that of the modified image from the input image; a learning image coefficient vector acquiring step of acquiring a coefficient vector of the learning image in the intermediate eigenspace; a weight coefficient determining step of determining a weight coefficient according to a mutual relationship between the coefficient vector of the learning image in the intermediate eigenspace and the coefficient vector of the input image in the intermediate eigenspace calculated by the first sub-tensor projecting step; and a synthesis step of determining an adopting ratio between a process including the first and second projective operations and a process by the high image quality processing step according to the weight coefficient, and combining the modified image and the high quality image.
-
-
28. An image processing method, including:
-
an eigen projective matrix generating step of generating an eigen projective matrix generated by a projective operation from a learning image group including pairs of first quality images and second quality images different in image quality from each other; a projective kernel tensor generating step of generating a projective kernel tensor specifying a corresponding relationship between the first quality images and an intermediate eigenspace and a corresponding relationship between the second quality images and the intermediate eigenspace; a first sub-kernel tensor acquiring step of generating a first sub-kernel tensor satisfying a condition specified by a first setting from the generated projective kernel tensor; a second sub-kernel tensor acquiring step of generating a second sub-kernel tensor satisfying a condition specified by a second setting from the generated projective kernel tensor; a first sub-tensor projecting step of projecting an input image as a processing target using a first projective operation utilizing the eigen projective matrix and the first sub-kernel tensor, and calculating a coefficient vector in the intermediate eigenspace; a second sub-tensor projecting step of generating a modified image having a different image quality from that of the input image by projecting the calculated coefficient vector using a second projective operation utilizing the second sub-kernel tensor and the eigen projective matrix; a high image quality processing step of generating a high quality image having the same size as that of the modified image from the input image; a learning image coefficient vector acquiring step of acquiring a coefficient vector of the learning image in the intermediate eigenspace; a weight coefficient determining step of determining a weight coefficient according to a mutual relationship between the coefficient vector of the learning image in the intermediate eigenspace and the coefficient vector of the input image in the intermediate eigenspace calculated by the first sub-tensor projection step; and a synthesis step of determining an adopting ratio between a process including the first and second projective operations and a process by the high image quality processing step according to the weight coefficient, and combining the modified image and the high quality image.
-
-
29. A non-transitory computer-readable medium in which computer readable code of an image processing program is stored, wherein the image processing program causes a computer to function as:
-
an information acquisition device for acquiring an eigen projective matrix generated by a projective operation from a learning image group including pairs of first quality images and second quality images different in image quality from each other, and a projective kernel tensor generated from the learning image group and the eigen projective matrix; a first sub-kernel tensor generation device for generating a first sub-kernel tensor satisfying a condition specified by a first setting from the acquired projective kernel tensor; a second sub-kernel tensor generation device for generating a second sub-kernel tensor satisfying a condition specified by a second setting from the acquired projective kernel tensor; a first sub-tensor projection device for projecting an input image as a processing target using a first projective operation utilizing the eigen projective matrix and the first sub-kernel tensor, and calculating a coefficient vector in an intermediate eigenspace; a second sub-tensor projection device for generating a modified image having a different image quality from that of the input image by projecting the calculated coefficient vector using a second projective operation utilizing the second sub-kernel tensor and the eigen projective matrix; a high image quality processing device for generating a high quality image having the same size as that of the modified image from the input image; a learning image coefficient vector acquisition device for acquiring a coefficient vector of the learning image in the intermediate eigenspace; a weight coefficient determination device for determining a weight coefficient according to a mutual relationship between the coefficient vector of the learning image in the intermediate eigenspace and the coefficient vector of the input image in the intermediate eigenspace calculated by the first sub-tensor projection device; and a synthesis device for determining an adopting ratio between a process including the first and second projective operations and a process by the high image quality processing device according to the weight coefficient, and combining the modified image and the high quality image.
-
-
30. A non-transitory computer-readable medium in which computer readable code of an image processing program is stored, wherein the image processing program causes a computer to function as:
-
an information acquisition device for acquiring an eigen projective matrix generated by a projective operation from a learning image group including pairs of first quality images and second quality images different in image quality from each other, a first sub-kernel tensor satisfying a condition specified by a first setting generated using a projective kernel tensor generated from the learning image group and the eigen projective matrix, and a second sub-kernel tensor satisfying a condition specified by a second setting generated using the projective kernel tensor; a first sub-tensor projection device for projecting an input image as a processing target using a first projective operation utilizing the eigen projective matrix and the first sub-kernel tensor, and calculating a coefficient vector in an intermediate eigenspace; a second sub-tensor projection device for generating a modified image having a different image quality from that of the input image by projecting the calculated coefficient vector using a second projective operation utilizing the second sub-kernel tensor and the eigen projective matrix; a high image quality processing device for generating a high quality image having the same size as that of the modified image from the input image; a learning image coefficient vector acquisition device for acquiring a coefficient vector of the learning image in the intermediate eigenspace; a weight coefficient determination device for determining a weight coefficient according to a mutual relationship between the coefficient vector of the learning image in the intermediate eigenspace and the coefficient vector of the input image in the intermediate eigenspace calculated by the first sub-tensor projection device; and a synthesis device for determining an adopting ratio between a process including the first and second projective operations and a process by the high image quality processing device according to the weight coefficient, and combining the modified image and the high quality image.
-
-
31. A non-transitory computer-readable medium in which computer readable code of an image processing program is stored, wherein the image processing program causes a computer to function as:
-
an eigen projective matrix generation device for generating an eigen projective matrix generated by a projective operation from a learning image group including pairs of first quality images and second quality images different in image quality from each other; a projective kernel tensor generation device for generating a projective kernel tensor specifying a corresponding relationship between the first quality images and an intermediate eigenspace and a corresponding relationship between the second quality images and the intermediate eigenspace; a first sub-kernel tensor acquisition device for generating a first sub-kernel tensor satisfying a condition specified by a first setting from the generated projective kernel tensor; a second sub-kernel tensor acquisition device for generating a second sub-kernel tensor satisfying a condition specified by a second setting from the generated projective kernel tensor; a first sub-tensor projection device for projecting an input image as a processing target using a first projective operation utilizing the eigen projective matrix and the first sub-kernel tensor, and calculating a coefficient vector in the intermediate eigenspace; a second sub-tensor projection device for generating a modified image having a different image quality from that of the input image by projecting the calculated coefficient vector using a second projective operation utilizing the second sub-kernel tensor and the eigen projective matrix; a high image quality processing device for generating a high quality image having the same size as that of the modified image from the input image; a learning image coefficient vector acquisition device for acquiring a coefficient vector of the learning image in the intermediate eigenspace; a weight coefficient determination device for determining a weight coefficient according to a mutual relationship between the coefficient vector of the learning image in the intermediate eigenspace and the coefficient vector of the input image in the intermediate eigenspace calculated by the first sub-tensor projection device; and a synthesis device for determining an adopting ratio between a process including the first and second projective operations and a process by the high image quality processing device according to the weight coefficient, and combining the modified image and the high quality image.
-
Specification