摘要 |
For frames sequentially functioning as a reference frame in a video sequence frame set, a motion estimate set is accessed. One motion estimate characterizes motion associated with pixels of each region of the reference frame in relation to regions of one frame of the set of frames, which is temporally displaced in time with respect to other frames in the set of frames. An additional motion estimate characterizes motion associated with pixels of each reference frame region in relation to a second frame of the set, which is temporally displaced from the one frame and other frames of the frame set. A temporal image prediction set, corresponding to the first and additional motion estimate, is predicted, based on an alignment of the reference frame regions over the frame set. The temporal image predictions are blended and a temporal predictor is generated over the frame set based on the blending. |
申请人 |
DOLBY LABORATORIES LICENSING CORPORATION;ORLICK, CHRISTOPHER, J.;WEBB, RICHARD;MILLER, SCOTT;SHIELDS, JEROME, D. |
发明人 |
ORLICK, CHRISTOPHER, J.;WEBB, RICHARD;MILLER, SCOTT;SHIELDS, JEROME, D. |