METHOD OF ADAPTIVE FRAME PREDICTION FOR MULTIVIEW VIDEO SEQUENCE CODING

摘要

Disclosed is a process of digital compression of multi-view video by the addition of data scene depth. Each frame of the multi-view video sequence is again encoded as non-overlapped blocks according to the predefined order of coding. When at least one encoded frame corresponding to the given view is detected, the encoded frame is designated as a reference frame; the synthesized frame for the reference frame is differentiated from the pixels of the non-overlapped blocks of the encoded frame; and a block that is spatially combined with the encoded block in the synthesized frame is designated as a virtual block according to the encoded frame. The position of the virtual block in the synthesized frame is determined according to the reference frame. [Reference numerals] (100) MVV encoding system; (101) Depth map encoding sub-system; (102) Video sequence encoding unit; (103) Additional video sequence encoding unit; (104) MVV decoding system; (AA) Input sequence; (BB) Input depth map; (CC) Decoded sequence; (DD) Decoded depth map