Edge smoothing block filtering and blending
View Patent ↗Aspects of an edge smoothing block filter and combinations of the filter with a motion compensated temporal filter are described. In one embodiment, edge smoothing block filtering includes selecting a current pixel to be filtered and selecting a candidate pixel within a search area about the current pixel. The edge smoothing block filtering generally seeks to identify candidate pixels having surrounding pixels similar to pixels surrounding the current pixel. The edge smoothing block filtering further computes a cost difference between pixels within a candidate pixel neighborhood and pixels within a current pixel neighborhood, and filters the current pixel based on the cost difference. Aspects of the filters and filter element combinations may preserve edges and textures adaptively based on image content. For example, diagonal or curved image edges may be filtered along edges while texture is preserved along the edges.
1. A method of edge smoothing block filtering, comprising:
selecting a current pixel as a pixel to be filtered;
selecting a candidate pixel within a search area about the current pixel;
computing a cost difference between pixels within a candidate pixel neighborhood about the candidate pixel and pixels within a current pixel neighborhood about the current pixel, wherein computing a cost difference comprises summing difference between each pixel within the candidate pixel neighborhood and a corresponding pixel within the current pixel neighborhood;
converting the cost difference to a similarity measure based on a weight function and a smoothing noise threshold, the smoothing noise threshold associated with at least one of a local contrast strength or noise register values, the smoothing noise threshold determined by a function of a fixed threshold value and a value determined using a plurality of cost differences, wherein the similarity measure indicates a measure of filter strength; and
filtering the current pixel based on the candidate pixel and the similarity measure, wherein filtering the current pixel comprises modifying the candidate pixel by a function of the similarity measure.
2. The method of claim 1 , wherein:
converting the cost difference to a similarity measure further comprises converting the cost difference to the similarity measure based on a weight function, the weight function being selected according to a measure of filter strength; and
the method further comprises setting a value of the smoothing noise threshold according to a function of the cost difference.
3. The method of claim 2 , wherein filtering the current pixel comprises filtering the current pixel based on a value of the candidate pixel weighted by the similarly measure.
4. The method of claim 1 , wherein computing a cost difference comprises, for each of a plurality of candidate pixels within the search area about the current pixel, computing a cost difference between pixels within a candidate pixel neighborhood about the candidate pixel and the pixels within the current pixel neighborhood about the current pixel.
5. The method of claim 4 , further comprising:
setting a value of the smoothing noise threshold based on a function of the cost differences of the plurality of candidate pixels, wherein
converting the cost difference to a similarity measure further comprises converting each of the cost differences to a similarity measure based on the smoothing noise threshold.
6. The method of claim 5 , wherein setting a value of the smoothing noise threshold further comprises setting the value of the smoothing noise threshold according to one of the fixed threshold value or a fraction of a maximum value of the cost differences.
7. The method of claim 6 , further comprising:
when the fraction of the maximum value of the cost differences is less than the fixed threshold value, setting the value of the smoothing noise threshold to the fixed threshold value; and
when the fraction of the maximum value of the cost differences is greater than or equal to the fixed threshold value, setting the value of the smoothing noise threshold to the fraction of the maximum value of the cost differences.
8. The method of claim 7 , further comprising setting the fraction of the fraction of the maximum value of the cost differences according to a desired amount of smoothing.
9. The method of claim 8 , further comprising normalizing each of the similarity measures.
10. The method of claim 9 , wherein filtering the current pixel comprises filtering the current pixel based on the value of each of the plurality of candidate pixels weighted by the normalized similarly measures.
11. An edge smoothing block filter, comprising:
a current pixel selector that selects a current pixel as a pixel to be filtered;
a candidate pixel selector that selects a candidate pixel within a search area about the current pixel;
a neighborhood calculator that computes a cost difference between pixels within a candidate pixel neighborhood about the candidate pixel and pixels within a current pixel neighborhood about the current pixel by summing difference between each pixel within the candidate pixel neighborhood and a corresponding pixel within the current pixel neighborhood;
a weighter that converts the cost difference to a similarity measure based on a weight function and a smoothing noise threshold, the weight function being selected according to a measure of filter strength and the smoothing noise threshold associated with at least one of a local contrast strength or noise register values, the smoothing noise threshold determined by a function of a fixed threshold value and a value determined using a plurality of cost differences; and
a current pixel updater that filters the current pixel based on a value of the candidate pixel weighted by the similarly measure.
12. The edge smoothing block filter of claim 11 , wherein the neighborhood calculator further computes, for each of a plurality of candidate pixels within the search area about the current pixel, a cost difference between pixels within a candidate pixel neighborhood about the candidate pixel and the pixels within the current pixel neighborhood about the current pixel.
13. The edge smoothing block filter of claim 12 , wherein the weighter sets a value of the smoothing noise threshold based on a function of the cost differences of the plurality of candidate pixels.
14. The edge smoothing block filter of claim 13 , wherein the weighter sets the value of the smoothing noise threshold according to one of the fixed threshold value or a fraction of a maximum value of the cost differences.
15. The edge smoothing block filter of claim 14 , wherein the weighter is configured to:
when the fraction of the maximum value of the cost differences is less than the fixed threshold value, set the value of the smoothing noise threshold to the fixed threshold value; and
when the fraction of the maximum value of the cost differences is greater than or equal to the fixed threshold value, set the value of the smoothing noise threshold to the fraction of the maximum value of the cost differences.