Video error concealment by integrating dynamic programming and adaptive Kalman filtering techniques

This paper addresses an error concealment technique, focused on motion vector recovery for P and B frames, to improve video quality at the decoder when video bit stream data incur transmission errors. We first proposed a dynamic programming (DP) technique to optimize the path cost in a multi-stage topology, which explores both the boundary matching errors and side smoothness between adjacent corrupted macroblocks. Due to the high computing complexity of DP, the sub-optimal alternative, enhanced with an adaptive Kalman filtering algorithm, is adopted instead. Experiments show that the proposed algorithm is capable of improving the reconstructed video quality by about 0.3/spl sim/1.0 dB at a packet loss rate of 10%, with respect to the traditional boundary matching algorithm. Besides, the subjective image inspection also shows the benefit of the proposed algorithm on retaining the continuity of lines and image details.