摘要：Multiple description coding (MDC) is a promising alternative to combatting information loss without any retransmission. In this paper, an effective MD video codec is designed based on temporal pre- and post-processing of video sequences without modifying the actual coding process itself, which makes it compatible with the current standard source or channel codec. For ease of post-processing, motion-compensated interpolation (MCI) based on piecewise uniform motion assumption is adopted to estimate the lost frame in side decoding. Accordingly, to match the post-processing, in the pre-processing joint frame duplication/interpolation is first applied to the original video data before performing odd/even frame splitting, which attempts to make the motion variety in the generated descriptions piecewise uniformly thus achieving better side reconstructed quality based on MCI. The experimental results exhibit better performance of the proposed scheme than some other tested schemes, in both the on-off channel environment and packet loss network.
Loading...
