Abstract
For real-time video streaming over unreliable networks, the losses or delay of the video packets may deteriorate the video quality at the receiver side. The traditional temporal error concealment methods try to guess the lost motion vectors (MVs) of the lost macroblocks (MBs) by exploiting the received motion information of the neighboring macroblocks, and that of the co-located macroblock in the previous frames. The accuracy of this motion vector will affect the performance of the error concealment method. In this paper, instead of only exploiting the motion information in the current and previous frames, we also use the motion information in the following frames. In this case, more accurate estimation of the motion vector could be achieved; This leads to better concealment performance. Experimental results demonstrate that the proposed method can improve the performance in comparison with the traditional error concealment methods, for instance, the gain could be up to 0.6 dB for the "Foreman" video sequence.