The Research About Error Concealment Of H.264Decoder Side

H.264video compression standard is widely used because of its efficient compression performance and good network of affinity, has become the mainstream of the current video coding standard. However, due to network congestion and other reasons, the loss of the compressed video data turn out to be a problem which is difficult to avoid, and the H.264standard has higher coding efficiency, at the same time, if the decoding error occurs, the error will spread in both spatial domain and time domain. And as data retransmission can not guarantee a good real-time video decoding, error concealment technology has become the last line of defense to ensure video quality, is highly valued and studied.This paper mainly studies the H.264decoder side error concealment techniques, and proposes two algorithms. The main research work and innovations are as follows:Firstly, after introducing the video compression standard and error concealment, further analysis about the principle of coder and decoder, and the key technologies of H.264video compression standard is given; after introducing error concealment techniques in video communication, further analysis about the error concealment in JM is given.Secondly, after analyzing the existing spatial error concealment, an adaptive spatial error concealment method based on direction information is proposed. Extract and use directional information to divide the lost macro blocks into smoothing blocks, single edge blocks, multi-edge blocks and complex texture blocks, and then select appropriate error concealment algorithm for each type of macro blocks to conceal them. H.264official test platform for simulation is used to test, results show that the algorithm of this paper achieves better visual quality than DI algorithm, with0-1.79dB gain in PSNR and0-0.88%gain in SSIM.Thirdly, after analyzing the paper adaptive spatial error concealment algorithm based on facial features, a new error concealment based on facial symmetry is proposed. First, detect face, and then judge the similarity to distinguish the image into symmetrical face region and other region. As for symmetrical face region, use symmetric concealment, for other region, use ASEC algorithm. And finally JM86test platform is used to test, results show that our algorithm get a better recovery effects than ASEC algorithm in the face region.