Limited-domain motion estimation/compensation for video encoding/decoding

摘要

Methods and apparatuses for encoding video images using motion estimation and decoding encoded video images using motion compensation, where the motion estimation and/or the motion compensation is limited to subimages of the video images as defined by boundaries. Although motion compensation may not be used to estimate pels using data from the previous image frame that lies across the subimage boundaries, motion compensation may be used to estimate a portion of a block of the image frame being decoded using a region from the previous image frame that abuts a subimage boundary and is smaller than the block. The rest of the block may be estimated either by retaining the corresponding pels from the previous image frame or by replicating the pels from the previous image frame that lie along the subimage boundary.