Study On Error Concealment Technology Based On H.264

With the rapid development of information age in human society, the application of video communication becomes a certain trend. But the tremendous data of video brings some difficulties for video transmission and storage, so it's necessary to use motion compensation, DCT and other technologies to compress the video data. When the compressed video stream is transmitted, it's easy to be corrupted, and the error may propagate to the succeed frames. On this condition, error concealment is an effective method to combat errors.In this paper, the H.264 standard is introduced in detail first, and then error concealment is researched particularly. The main study of this paper concludes two parts: I-frame error concealment and P-frame error concealment. For I-frame error concealment, since the traditional linear interpolation has good effect in smooth video picture while misses the edge information, an improved I-frame error concealment method combining the direction interpolation and boundary matching is proposed in this paper. For an error block of I-frame, the proposed algorithm first calculates the average direction of each neighboring block which is decoded correctly, and uses the direction interpolation method to rebuild a corresponding predictive block. Then the block which has minimum error is chose as the last concealed block from the predictive blocks using the boundary matching. For P-frame error concealment, since the zero motion vector algorithm is not suitable for the video picture with rapid motion, and the boundary matching algorithm can't estimate motion vector accurately, an improved P-frame error concealment algorithm is proposed. For an error block, both the average motion vector and the medial motion vector of all neighboring blocks which are decoded correctly are added to the optional motion vectors. And the boundary matching principle is amended correspondingly. Finally a more accurate motion vector is searched.Verifying these two schemes on JM model which is the standard test platform of H.264, the experimental results show that the proposed algorithms obtain better performance both in subjective and objective effect than JM.