System and method for filtering an image
View Patent ↗An embodiment is configured to calculate a perceptual masking factor at a pixel location at a block boundary of the image, calculate a parameter for a filter at the pixel location at the block boundary, and filter the image around the pixel location at the block boundary employing the filter with the calculated parameter. The perceptual masking factor is formed as a product of a background activity masking map and a luminance masking map. The filter includes a parameter that is selected in view of a quality of experience performance measure for the image at the pixel location at the block boundary of the image.
1. An apparatus configured to filter an image, comprising:
a memory; and
a processor coupled to the memory, the processor, in conjunction with the memory, configured to, separately for each pixel location at a horizontal or vertical block boundary of the image:
calculate a perceptual masking factor at the pixel location at the block boundary of the image,
calculate a parameter for a filter at the pixel location at the block boundary by minimizing a regularized cost function for the parameter, and
filter the image around the pixel location at the block boundary employing the filter with the calculated parameter.
2. The apparatus as in claim 1 wherein the perceptual masking factor comprises a product of a background activity masking map and a luminance masking map.
3. The apparatus as in claim 1 wherein the parameter is calculated for each pixel location in view of a quality of experience performance measure for the image at the pixel location at the block boundary of the image.
4. The apparatus as in claim 3 wherein the parameter is calculated for each pixel location to maximize the quality of experience performance measure for the image at the pixel location at the block boundary of the image.
5. The apparatus as in claim 3 wherein the parameter is calculated for each pixel location in view of the quality of experience performance measure for the image employing a search over a limited range of values for the parameter.
6. The apparatus as in claim 3 wherein the quality of experience performance measure comprises a summation of image slopes at block boundaries weighted by the perceptual masking factor.
7. The apparatus as in claim 1 wherein the parameter comprises a standard deviation parameter.
8. The apparatus as in claim 1 wherein the filter comprises a Gaussian filter.
9. The apparatus as in claim 1 wherein the block boundary is the horizontal block boundary.
10. The apparatus as in claim 9 wherein the processor, in conjunction with the memory, is further configured to, separately for each pixel location at the vertical block boundary of the image:
calculate the perceptual masking factor at the pixel location at the vertical block boundary of the image;
calculate the parameter for the filter at the pixel location at the vertical block boundary by minimizing the regularized cost function for the parameter; and
filter the image around the pixel location at the vertical block boundary employing the filter with the calculated parameter.
11. The apparatus as in claim 1 wherein the processor, in conjunction with the memory, is further configured to filter the image at pixel locations to a left and to a right of the block boundary employing the filter with the calculated parameter.
12. A method of filtering an image, the method comprising:
separately for each pixel location at a horizontal or vertical block boundary of the image:
calculating, by a processor, a perceptual masking factor at the pixel location at the block boundary of the image;
calculating, by the processor, a parameter for a filter at the pixel location at the block boundary by minimizing a regularized cost function for the parameter; and
filtering, by the processor, the image around the pixel location at the block boundary employing the filter with the calculated parameter.
13. The method as recited in claim 12 wherein the perceptual masking factor comprises a product of a background activity masking map and a luminance masking map.
14. The method as recited in claim 12 wherein the parameter is calculated for each pixel location in view of a quality of experience performance measure for the image at the pixel location at the block boundary of the image.
15. The method as recited in claim 14 wherein the quality of experience performance measure comprises a summation of image slopes at block boundaries weighted by the perceptual masking factor.
16. The method as recited in claim 12 wherein the parameter comprises a standard deviation parameter.
17. The method as recited in claim 12 wherein the filter comprises a Gaussian filter.
18. The method as recited in claim 12 wherein the block boundary is the horizontal block boundary.
19. The method as recited in claim 18 further comprising:
separately for each pixel location at the vertical block boundary of the image:
calculating the perceptual masking factor at the pixel location at the vertical block boundary of the image;
calculating the parameter for the filter at the pixel location at the vertical block boundary by minimizing the regularized cost function for the parameter; and
filtering the image around the pixel location at the vertical block boundary employing the filter with the calculated parameter.
20. The method as in claim 12 further comprising filtering the image at pixel locations to a left and to a right of the block boundary employing the filter with the calculated parameter.
21. The method as recited in claim 12 further comprising calculating the parameter for each pixel location using a local gradient of the regularized cost function.
22. The apparatus as in claim 1 wherein the processor is configured to calculate the parameter for each pixel location using a local gradient of the regularized cost function.