Optimization And Implementation Of Wyner-Ziv Distributed Video Coding System

With the development of wireless video sensor network, mobile video communication and other new applications, traditional video coding technology has been difficult to adapt to these new applications because of high encoding complexity, and distributed video coding came into being. Distributed video coding is a new coding scheme. The main feature of distributed video coding is to transfer the complexity from encoder to decoder. It is of great significance to carry out the research on distributed video coding for video transmission of field military communication, emergency rescue and other special occasions.Based on the theory of distributed video coding, this thesis studies the Wyner-Ziv video coding scheme, also designs and develops an optimized pixel-domain Wyner-Ziv video coding prototype system. The main research contents and contributions are listed as follows:(1) This thesis designs the pixel-domain Wyner-Ziv video coding system framework based on LDPC to solve the error-floor effect of Turbo, and simulates the system. The system based on LDPC can get better performance compared to the system based on Turbo, thus proving that LDPC code has better error correction capability than Turbo code. The system based on LDPC can improve the robustness of video transmission.(2) This thesis proposes an improved multi-path side information generation strategy based on the comparison analysis of the typical side information generation methods. This strategy can get better side information, thus reducing the transmission of parity bits from the encoder and improving the performance and efficiency of the system.(3) In view of the reality that key frames may occur transmission error in the practical application of distributed video coding, this thesis proposes using error concealment technique to process the decoded key frames, thus ensuring the transmission quality of the key frames and improving the robustness of system. Then we use concealment technique to process the decoded key frame and generate side information by the proposed side information generation strategy. We can get better side information, thus improving reconstruction quality and the performance of system, so that we can optimize the Wyner-Ziv system.