A temporal error concealment algorithm based on structural alignment for packet video

Abstract

With rapid growth of packet switched communication using the IP protocol, concealment of packet-loss related distortions in multimedia communications assumes great importance. This paper proposes a novel temporal concealment algorithm to conceal packet-loss related distortion in video transmission. The algorithm uses the previous frame (assumed to be error-free) to conceal the distortion in the present frame. It is based on the insight that the spatial region around the area in the previous frame that yields the best concealment will in general be highly correlated with the corresponding area around the region to be concealed in the present frame. We implement the proposed algorithm for a MPEG-2 based packet-loss scenario and compare its performance with prevalent algorithms based on boundary matching. The proposed algorithm effects better concealment as compared to these algorithms. In particular, it exhibits a superior ability to align edges. With its low computation complexity, this algorithm can find application in real-time concealment of packet-loss related artifacts in streaming video.