Video processing with region-based warping

1. A method for processing video data, the method comprising, for each of a plurality of source video frames:

using pixels representing the source video frame and at least one temporally-adjacent video frame, generating a motion history map based upon frame-to-frame movement of at least one region in the source video frame and temporally-adjacent video frames;

identifying at least one salient region in the source video frame based upon the motion history map and static image saliency data;

determining weighting data for warping pixels corresponding to background regions of the source video frame differently than pixels corresponding to the at least one salient region in the source video frame; and

using the weighting data, warping the pixels representing the source video frame to generate pixels representing a warped video frame.

2. The method of claim 1 , wherein identifying at least one salient region in the source video frame based upon the motion history map and static image saliency data includes

identifying at least one image-salient region in the source video frame based upon static image saliency data that identifies at least one salient region in the source video frame, and

identifying salient regions including the identified at least one image-salient region and at least one region exhibiting frame-to-frame movement in the motion history map.

3. The method of claim 1 , wherein identifying at least one salient region in the source video frame based upon the motion history map and static image saliency data includes

in response to the static image saliency identifying no image-salient regions, identifying the at least one salient region based upon at least one region exhibiting frame-to-frame movement in the motion history map.

4. The method of claim 1 , wherein warping the pixels representing the source video frame to generate pixels representing a warped video frame includes

determining a current warp function by generating a modified warp function based upon a previous warp function applied to a video frame in the video data that temporally precedes the source video frame, using the weighting data to constrain modifications to the warp function to be applied to pixels corresponding to the background region, relative to the warp function to be applied to pixels corresponding to the salient regions, and

applying the current warp function to the source video frame to warp the pixels representing the source video frame to resize at least the background region.

5. The method of claim 4 , wherein determining a current warp function by generating a modified warp function based upon a previous warp function includes modifying the previous warp function.

6. The method of claim 4 , wherein determining a current warp function by generating a modified warp function based upon a previous warp function includes determining a current warp function for warping the warped video frame, and modifying the current warp function based upon the previous warp function.

7. The method of claim 4 , wherein using the weighting data to constrain modifications to the warp function to be applied to pixels corresponding to the background region, relative to the warp function to be applied to pixels corresponding to the salient regions, includes constraining modifications to the warp function to be applied to the salient region pixels to preserve both the shape of the salient region pixels and the continuity of the warp function as applied to the salient region pixels in at least one adjacent frame.

8. The method of claim 4 , wherein applying the current warp function includes resizing the salient regions.

9. The method of claim 1 , wherein generating a motion history map based upon frame-to-frame movement of at least one region in the source in temporally-adjacent video frames includes identifying an area in the source video frame, that spatially covers areas in which the at least one region is located in the temporally-adjacent video frames.

10. The method of claim 1 ,

further including generating an image saliency map for the source video frame based upon static image-saliency data for a plurality of video frames temporally adjacent the source video frame,

wherein identifying at least one salient region in the source video frame based upon the motion history map and static image saliency data includes identifying at least one salient region based on the image saliency map.

11. The method of claim 1 , wherein generating a motion history map based upon frame-to-frame movement of at least one region in the source in temporally-adjacent video frames includes blending motion saliency data for each of the source video frame and plurality of temporally-adjacent video frames.

12. The method of claim 1 ,

further including setting a number of temporally-adjacent video frames for use in generating the motion history map based upon a tradeoff condition between temporal coherence and shape preservation for salient regions in the source video frames, and

wherein generating the motion history map based upon frame-to-frame movement of at least one region in the source video frame and temporally-adjacent video frames includes generating the motion history map based upon frame-to-frame movement of the at least one region in the set number of temporally-adjacent video frames.

13. A method for processing a stream of video data, the method comprising:

in response to a user input setting a tradeoff condition between temporal coherence and shape preservation, setting a weighting parameter for applying a warp to each of a plurality of video frames in the stream of video data;

for each of the video frames, determining at least one region of the frame as a salient region based upon

frame-to-frame movement of an object in a set number of video frames that are temporally adjacent the video frame, and

image saliency data that identifies image-salient regions in the video frame; and

using the set weighting parameter, weighting, for each of the plurality of video frames, background regions of the image differently from the at least one region of the frame determined as salient.

14. The method of claim 13 , further including, for each of the plurality of video frames, resizing the video frame using the weighting for the frame to resize the background regions differently from the at least one salient region determined for the frame.

15. The method of claim 13 , wherein setting a weighting parameter includes setting a number of temporally-adjacent video frames to use in determining frame-to-frame movement of at least one region in the temporally-adjacent video frames.

16. An article of manufacture, comprising:

a non-transitory processor-readable storage medium configured with configuration data that when executed by a processor, cause the processor to, for each of a plurality of source video frames:

generate a motion history map based upon frame-to-frame movement of at least one region in the source video frame and temporally-adjacent video frames using pixels representing the source video frame and at least one temporally-adjacent video frame,

identify at least one salient region in the source video frame based upon the motion history map and static image saliency data,

determine weighting data for warping pixels corresponding to background regions of the source video frame differently than pixels corresponding to the at least one salient region in the source video frame, and

using the weighting data, warp the pixels representing the source video frame to generate pixels representing a warped video frame.

17. The article of manufacture of claim 16 , wherein the processor-readable storage medium is further configured with configuration data that when executed by a processor, cause the processor to, for each of a plurality of source video frames, identify at least one salient region in the source video frame based upon the motion history map and static image saliency data by

identifying at least one image-salient region in the source video frame based upon static image saliency data that identifies at least one salient region in the source video frame, and

identifying salient regions including the identified at least one image-salient region and at least one region exhibiting frame-to-frame movement in the motion history map.

18. The article of manufacture of claim 16 , wherein the processor-readable storage medium is further configured with configuration data that when executed by a processor, cause the processor to, for each of a plurality of source video frames, identify at least one salient region in the source video frame based upon the motion history map and static image saliency data by, in response to the static image saliency identifying no image-salient regions, identifying the at least one salient region based upon at least one region exhibiting frame-to-frame movement in the motion history map.

19. The article of manufacture of claim 16 , wherein the processor-readable storage medium is further configured with configuration data that when executed by a processor, cause the processor to, for each of a plurality of source video frames, warp the pixels representing the source video frame to generate pixels representing a warped video frame by

determining a current warp function by generating a modified warp function based upon a previous warp function applied to a video frame that temporally precedes the source video frame, using the weighting data to constrain modifications to the warp function to be applied to pixels corresponding to the background region, relative to the warp function to be applied to pixels corresponding to the salient regions, and

applying the current warp function to the source video frame to warp the pixels representing the source video frame.

20. The article of manufacture of claim 19 , wherein determining a current warp function by generating a modified warp function based upon a previous warp function includes modifying the previous warp function.

21. The article of manufacture of claim 19 , wherein determining a current warp function by generating a modified warp function based upon a previous warp function includes determining a current warp function for warping the warped video frame, and modifying the current warp function based upon the previous warp function.

22. The article of manufacture of claim 16 , wherein the processor-readable storage medium is further configured with configuration data that when executed by a processor, cause the processor to, for each of a plurality of source video frames,

generate an image saliency map for the source video frame based upon static image-saliency data for a plurality of video frames temporally adjacent the source video frame, and

identify at least one salient region in the source video frame based upon the motion history map and static image saliency data, by identifying at least one salient region based on the image saliency map.

23. The article of manufacture of claim 16 , wherein the processor-readable storage medium is further configured with configuration data that when executed by a processor, cause the processor to, for each of a plurality of source video frames,

set a number of temporally-adjacent video frames for use in generating the motion history map based upon a tradeoff condition between temporal coherence and shape preservation for salient regions in the source video frames, and

generate the motion history map based upon frame-to-frame movement of the at least one region in the set number of temporally-adjacent video frames.

24. An apparatus for processing video data, the apparatus comprising:

a logic circuit configured to, for each of a plurality of source video frames

generate a motion history map based upon frame-to-frame movement of at least one region in the source video frame and temporally-adjacent video frames using pixels representing the source video frame and at least one temporally-adjacent video frame,

identify at least one salient region in the source video frame based upon the motion history map and static image saliency data,

determine weighting data for warping pixels corresponding to background regions of the source video frame differently than pixels corresponding to the at least one salient region in the source video frame, and

using the weighting data, warp the pixels representing the source video frame to generate pixels representing a warped video frame.

25. The apparatus of claim 24 , wherein the logic circuit is configured to

determine the weighting data by, in response to a user input setting a tradeoff condition between temporal coherence and shape preservation, setting a weighting parameter for applying a warp to each of a plurality of video frames in the video data, and

warp the pixels by using the weighting parameter to warp the pixels representing the source video frame to generate the pixels representing the warped video frame.