Saliency based denoising
AssigneeGoogle Inc.
InventorsMohammad Izadi, Yeping Su, Jeremy Dorfman
Life of this patent
Estimating the term…
Abstract
Image denoising includes obtaining a saliency map for an image. The saliency map includes respective saliency scores for pixels of the image. Respective noise levels are assigned to the pixels using the respective saliency scores to obtain a noise level map. The image is denoised using the noise level map to obtain a denoised image. The denoised image is output, such as to a display or a storage device.
Claims20
A method for image denoising, comprising:
- obtaining a saliency map for an image, wherein the saliency map comprises respective saliency scores for pixels of the image, wherein the respective saliency scores are indicative of perceptual importance of the pixels relative to each other, and wherein the respective saliency scores are normalized to a range of [0, 1];
- assigning respective noise levels to the pixels using the respective saliency scores to obtain a noise level map;
- sorting the respective saliency scores of the noise level map to obtain a sorted list of saliency scores,
- dividing the saliency scores of the sorted list of saliency scores into a predetermined number of saliency score groups,
- generating groups of pixels corresponding to the saliency score groups;
- assigning respective noise levels to the groups of pixels, wherein a respective noise level is assigned to each pixel of a group of pixels;
- denoising the image using the noise level map to obtain a denoised image;
- and
- outputting the denoised image.
A device for image denoising, comprising:
- a processor configured to:
- obtain a saliency map for an image, wherein the saliency map comprises respective saliency scores for pixels of the image, wherein the respective saliency scores are indicative of perceptual importance of the pixels relative to each other, and wherein the respective saliency scores are normalized to a range of [0, 1];
- assign respective noise levels to the pixels using the respective saliency scores to obtain a noise level map;
- sort the respective saliency scores of the noise level map to obtain a sorted list of saliency scores;
- divide the saliency scores of the sorted list of saliency scores into a predetermined number of saliency score groups;
- generate groups of pixels corresponding to the saliency score groups;
- assign respective noise levels to the groups of pixels, wherein a respective noise level is assigned to each pixel of a group of pixels;
- denoise the image using the noise level map to obtain a denoised image;
- and
- output the denoised image.
A non-transitory computer readable medium storing instructions operable to cause one or more processors to perform operations for image denoising comprising:
- obtaining a saliency map for an image, wherein the saliency map comprises respective saliency scores for pixels of the image, wherein the respective saliency scores are indicative of perceptual importance of the pixels relative to each other, and wherein the respective saliency scores are normalized to a range of [0, 1];
- assigning respective noise levels to the pixels using the respective saliency scores to obtain a noise level map;
- sorting the respective saliency scores of the noise level map to obtain a sorted list of saliency scores;
- dividing the saliency scores of the sorted list of saliency scores into a predetermined number of saliency score groups;
- generating groups of pixels corresponding to the saliency score groups,
- assigning respective noise levels to the groups of pixels, wherein a respective noise level is assigned to each pixel of a group of pixels;
- denoising the image using the noise level map to obtain a denoised image;
- and
- outputting the denoised image.
Dependencies are inferred from the claim text.