Research Of Error Concealment Technology Based On H.264

In the past 20 years, multimedia communication was developed rapidly worldwide, and was introduced to new video-information businesses constantly. In coding algorithm of video, a series of standards have been already put out. The latest standard is H.264/AVC that was established finally in May, 2003. Through inheriting, ameliorating and innovating, H.264 synthesizes so many advantages of the previous standards, which not only enhances its compress performance, but also gets eminent predominance in network friendship and error resilience. But it is impossible to avoid transmission error as interruption and delay always happen in channel. Since highly compressed video information is very sensitive to interference, error resilience becomes significant in video transmission.In this thesis, we first introduce basic principles of digital video compression coding and several aspects of error control. Then based on the analysis of video coding standard H.264, the error control and concealment technologies with the overall system analysis in the H.264/AVC standard are discussed. At last, this paper focuses on the post processing technique which is known as error concealment. Based on the research of many correlated work domestic and foreign, we analyze some error concealment algorithms of H.264 video coding standards in details and points out problems of each algorithm. After that we proposed our refined algorithms in both spatial and temporal areas.In spatial area, we proposed a multi-directional interpolation algorithm which can adaptive divide missing MB into several areas to restore. Experimental results showed that our algorithm can not only achieve good directional interpolation concealment results in both visual effect and PSNR value but also can avoid fake edges.In temporal area, we propose an effective error concealment algorithm base on variable MB mode. Depending on the modes of neighboring MB, each lost MB is concealed on the basis of different block sizes. Simulation results show the proposed method yields better video quality than conventional approaches.