Study On Error Control For Video Transmission Over Error-Prone Channels

The recent advancement of communication and video compression technology made digital video transmission a reality. Current wire and wireless transmission channel cannot guarantee reliable delivery of compressed video bit-stream. Bit errors or packet losses may occur during transmission. Most video coding standards adopt block-based hybrid coding algorithms to reduce the temporal redundancy by inter frame prediction, reduce spatial redundancy of remaining signal by transform coding and reduce statistic redundancy by variable length coding. While the hybrid coding algorithms achieve high coding efficiency, the compressed video bit-stream is more vulnerable to transmission errors. One bit error may result in failure of decoding of all the bits between two synchronization markers and packet loss may lead to continuously erroneous microblocks in a frame. Moreover, the errors in one frame may propagate to successive frames by motion compensation. High efficiency and robust video coding and transmission algorithms should be developed to tradeoff the coding efficiency and anti-transmission error capability. Error concealment algorithms also should be implemented at video decoder to reduce the quality degradation caused by transmission errors.At the video encoder, an intra refresh scheme is proposed for wireless video transmission to stop inter frame error propagation. Two-state Markov model is adopted to simulate the fading wireless channel. The error propagation processes of pixel level and microblock level are analyzed combined with specific error concealment algorithms at decoder. The propagation distortion is incorporated into rate-distortion frame to develop a optimized intra refresh scheme. The location and number of the intra microblocks are determined efficiently to stop inter frame error propagation based on transmission states of wireless channel and error propagation. Feedback information is also exploited to enhance the robustness of wireless channel model and error propagation analysis. The proposed intra refresh scheme is compatible with video coding standards and can be applied to real time end to end wireless video communication.Joint Source/Channel rate decision is studied and a forward error correction (FEC) coding rate decision algorithm for video transmitted over Rayleigh fading wireless channel is proposed. The objective is to decide optimal FEC code rate and minimize the decoder distortion. Residual packet loss rate of packet level FEC of wireless channel is analyzed. The source distortion is estimated by H.263 rate-distortion model. The channel distortion is estimated with the error concealment algorithms and channel transmission conditions. By taking consideration of source distortion and channel distortion together, the optimization problem for FEC code rate is resolved. The joint rate decision can allocate bandwidth to video source encoder and channel FEC fairly at limited transmission bandwidth and can enhance video transmission quality over wireless channel.Error concealment is studied in this dissertation. A temporal error concealment and a spatial error concealment algorithm based on edge recovery are proposed. Firstly, the spatial correlation of erroneous microblock is analyzed and a reliable edge pixel detection, linkage and recovery algorithm is proposed. In the proposed spatial error concealment algorithm, recovered edges divide the erroneous block into several sub-blocks with smooth pixel value distribution. Directional interpolation is applied to each smooth sub-block to recover the lost pixels. In the proposed temporal error concealment algorithm, the recovered edges separate the edged microblock into several sub-blocks with homogenous motion activity. Each sub-block is concealed by decoder motion estimation in the reference frame using correctly decoded boundary pixels and recovered edge pixels as reference pixels. Dividing the microblock by edge makes the sub-block satisfy smooth pixel value distribution and motion consistent better. The proposed spatial and temporal error concealment algorithms enable the decoder achieve better image recovery quality. The proposed error concealment algorithms don't increase the transmission rate and can be implemented in most video transmission system.At the end, the research works of this dissertation are summarized and further researches in this field are proposed.