| 摘要 |
A method for performing decoder-side motion estimation comprises the steps of (i) determining a first motion vector MVO in a search path which points from a current block to a reference block within a specified search window and (ii) determining a second motion vector MVO pointing to a reference block in a second reference frame where MV1 is a function of MV0. A combined metric for each MV0,MV1 pair in the search path is determined, wherein the metric is combination of a first metric based on temporal frame correlation, a second metric based on spatial neighbours of the reference blocks and a third metric based on the spatial neighbours of the current block. Thereafter, the MV0 with a desirable combined metric is selected and used as a motion vector for the current block. By performing the motion estimation at the decoder side bandwidth requirements for video data transmission may be reduced. |
