Scale-space self-similarity image processing
AssigneeHewlett-Packard Development Company, L.P.
InventorRenato Keshet
Life of this patent
Estimating the term…
Abstract
Image processing includes computing image ft from image fs, where images ft and fs are representations of a self-similar image at scales t and s in a scale-space, where t<s. Such processing may adapted for image deblurring, denoising and upscaling.
Claims44
An image processing method, comprising:
- computing image ft from image fs, wherein the images ft and fs respectively are representations of a self-similar image f0 at scales t and s in a scale-space, t<s, the image ft is upstream of the image fs in the scale-space and thereby closer in the scale-space to the self-similar image f0 than the image fs, and the scale-space is defined by a family of operators {T(k)} each of which maps the self-similar image f0 to a respective representation of the self-similar image f0 at a scale k;
- wherein the computing comprises determining one or more parameter values specifying at least one of the operators {T(k)} based on values of the image fs;
- and
- wherein the computing is performed by a processor.
A computer-readable medium comprising computer-readable instructions that, when executed by a processor, cause the processor to perform operations comprising:
- accessing an input digital image fs;
- and
- computing an output digital image ft from the input digital image fs, wherein the output and input images respectively are representations of a self-similar image f0 at scales t and s in a scale-space, t<s, the image ft is upstream of the input image fs in the scale-space and thereby closer in the scale-space to self-similar image f0 than the input fs, and the scale-space is defined by a family of operators {T(k)} each of which maps the self-similar image f0 to a respective representation of the self-similar image f0 at a scale k;
- wherein the computing comprises determining one or more parameter values specifying at least one of the operators {T(k)} based on values of the image fs.
Apparatus, comprising:
- a computer-readable medium storing computer-readable instructions;
- and
- a processor coupled to the computer-readable medium, operable to execute the instructions, and based at least in part on the execution of the instructions operable to perform operations comprising
- computing digital image ft from digital image fs, wherein the images ft and fs respectively are representations of a self-similar image f0 at scales t and s in a scale-space, t<s, the image ft is upstream of the input image fs in the scale-space and thereby closer in scale-space to self-similar image f0 than the image fs, and the scale-space is defined by a family of operators {T(k)} each of which maps the self-similar image f0 to a respective representation of the self-similar image f0 at a scale k;
- wherein the computing comprises determining one or more parameter values specifying at least one of the operators {T(k)} based on values of the image fs.
A computer-readable medium comprising computer-readable instructions that, when executed by a processor, cause the processor to perform operations comprising:
- computing digital image ft from digital image fs, wherein the images ft and fs respectively are representations of a self-similar image f0 at scales t and s in a scale-space, t<s, the image ft is upstream of the image fs in the scale-space and thereby closer in the scale-space to self-similar image f0 than the image fs, and the scale-space is defined by a family of operators {T(k)} each of which maps the self-similar image f0 to a respective representation of the self-similar image f0 at a scale k;
- wherein the computing comprises determining one or more parameter values specifying at least one of the operators {T(k)} based on values of the image fs.
Dependencies are inferred from the claim text.