Efficient region of interest detection

1. A video encoding apparatus comprising: a processor to:

receive a video signal;

compress a current picture in the video signal;

calculate a plurality of complexity values for all blocks of said current picture, wherein each of said complexity values characterizes how a corresponding one of a plurality of blocks in said current picture was compressed;

identify one or more of the blocks that have complexity values satisfying a discrimination threshold as candidate blocks associated with said current picture, wherein the discrimination threshold is based on at least one of an average of the plurality of complexity values, or a function of a maximum of the complexity values;

identify one or more of the candidate blocks that have complexity values satisfying a decision threshold as included within an initial ROI, wherein the decision threshold is based on an average of the complexity values of only the candidate blocks;

move a boundary of the initial ROI toward blocks with complexity values satisfying an adjustment threshold to define a final ROI, wherein the adjustment threshold is based on a local minimum of complexity values from one or more of the candidate blocks adjacent to a boundary of the initial ROI and excluded from the initial ROI;

superimpose said final ROI on a next picture in said video signal; and

compress said next picture with a higher bit rate inside said final ROI than outside said final ROI.

2. The apparatus according to claim 1 , wherein said processor is further to:

weight the plurality of complexity values toward a center of said current picture, and then identify the one or more blocks as candidate blocks based on the weighted complexity values and the discrimination threshold; or

reduce a dynamic range of the candidate block complexity values by applying a logarithmic function to the candidate block complexity values.

3. The apparatus according to claim 1 , wherein said processor is further to;

determine the discrimination threshold based on at least one of the average of the plurality of complexity values, or the function of a maximum of the complexity values;

determine the decision threshold based on the average of the candidate block complexity values; and

determine the adjustment threshold based on the local minimum of complexity values from the candidate blocks adjacent to the initial ROI boundary.

4. The apparatus according to claim 1 , wherein said processor is further to expand said initial ROI to include QxP adjacent blocks around each of the candidate blocks satisfying the decision threshold, wherein Q and P range from 2 to 4 adjacent blocks.

5. The apparatus according to claim 1 , wherein said processor is further to search in a plurality of directions from the boundary of the initial ROI to find one or more high complexity values within the initial ROI; and

terminate the search of a current row of blocks in response to finding a high complexity block satisfying the adjustment threshold.

6. The apparatus according to claim 5 , wherein said processor is further to move the boundary to the high complexity block satisfying the adjustment threshold to define the final ROI.

7. A method of video encoding, comprising:

receiving a video signal;

compressing a current picture in the video signal;

calculating a plurality of complexity values for all blocks of said current picture, wherein each of said complexity values characterizes how a corresponding one of a plurality of blocks in said current picture was compressed;

identifying one or more of the blocks that have complexity values satisfying a discrimination threshold as candidate blocks associated with said current picture, wherein the discrimination threshold is based on at least one of an average of the plurality of complexity values, or a function of a maximum of the complexity values;

identifying one or more of the candidate blocks that have complexity values satisfying a decision threshold as included within an initial ROI, wherein the decision threshold is based on an average of the complexity values of only the candidate blocks;

moving a boundary of the initial ROI toward blocks with complexity values satisfying an adjustment threshold to define a final ROI, wherein the adjustment threshold is based on a local minimum of complexity values from one or more of the candidate blocks adjacent to a boundary of the initial ROI and excluded from the initial ROI;

superimposing the final ROI on a next picture in said video signal; and

compressing said next picture with a higher bit rate inside said final ROI than outside said final ROI.

8. The method of claim 7 , further comprising:

weighting the plurality of complexity values toward a center of said current picture, and then identify the one or more blocks as candidate blocks based on the weighted complexity values and the discrimination threshold; or

reducing a dynamic range of the candidate block complexity values by applying a logarithmic function to the candidate block complexity values.

9. The method of claim 7 , further comprising:

determining the discrimination threshold based on at least one of the average of the plurality of complexity values, or the function of a maximum of the complexity values;

determining the decision threshold based on the average of the candidate block complexity values; and

determine the adjustment threshold based on the local minimum of complexity values from the candidate blocks adjacent to the initial ROI boundary.

10. The method of claim 7 , further comprising:

expanding said initial ROI to include QxP adjacent blocks around each of the candidate blocks satisfying the decision threshold, wherein Q and P each range from 2 to 4 adjacent blocks.

11. The method of claim 7 , further comprising:

searching in a plurality of directions from the boundary of the initial ROI to find one or more high complexity values within the initial ROI; and

terminating the search of a current row of blocks in response to finding a high complexity block satisfying the adjustment threshold.

12. The method of claim 11 , further comprising:

moving the boundary to the high complexity block satisfying the adjustment threshold to define the final ROI.

13. A non-transitory computer readable media, with instructions stored thereon, which when executed by a processor, cause the processor to perform a method comprising:

receiving a video signal;

compressing a current picture in the video signal;

calculating a plurality of complexity values for all blocks of said current picture, wherein each of said complexity values characterizes how a corresponding one of a plurality of blocks in said current picture was compressed;

identifying one or more of the blocks that have complexity values satisfying a discrimination threshold as candidate blocks associated with said current picture, wherein the discrimination threshold is based on at least one of an average of the plurality of complexity values, or a function of a maximum of the complexity values;

identifying one or more of the candidate blocks that have complexity values satisfying a decision threshold as included within an initial ROI, wherein the decision threshold is based on an average of the complexity values of only the candidate blocks;

moving a boundary of the initial ROI toward blocks with complexity values satisfying an adjustment threshold to define a final ROI, wherein the adjustment threshold is based on a local minimum of complexity values from one or more of the candidate blocks adjacent to a boundary of the initial ROI and excluded from the initial ROI;

superimposing the final ROI on a next picture in said video signal; and

compressing said next picture with a higher bit rate inside said final ROI than outside said final ROI.

14. The apparatus according to claim 1 , wherein:

the adjustment threshold is 200-400% of a minimum complexity value for candidate blocks to the left, right, above, or below the initial ROI boundary; and

the processor is to search in multiple directions, starting at the boundary of the initial ROI and moving toward the center of the initial ROI, for one or more blocks within the initial ROI that have higher complexity values than the adjustment threshold.

15. The apparatus according to claim 1 , wherein the discrimination threshold is a function of a maximum complexity value divided by a number between approximately 3 and 5.

16. The method of claim 7 , wherein:

the adjustment threshold is 200-400% of a minimum complexity value for candidate blocks to the left, right, above, or below the initial ROI boundary; and

moving the boundary further comprises searching in multiple directions, starting at the boundary of the initial ROI and moving toward the center of the initial ROI, for one or more blocks within the initial ROI that have higher complexity values than the adjustment threshold.

17. The method of claim 7 , wherein, wherein the discrimination threshold is a function of the maximum complexity value divided by a number between approximately 3 and 5.