Multiple Description Video Coding Based On H.264 Coding Standard

Multiple description coding (MDC) is an effective means to improve the robustness of video streams to packet losses and data errors in the Internet and wireless networks. H.264/AVC is the latest video coding standard, not only with high compression efficiency, but also offers a variety of novel error resilience coding tools, the standard is playing an increasingly important role in the network multimedia applications. How to effectively use the error resilience Mechanism of H.264/AVC standard, combined with multiple description coding methods to further improve the robustness of video streams to packet losses and data errors is of certain theoretical and practical significance.This thesis reviews the development of multiple description video coding, concentrate its study on the error resilience coding schemes of H.264/AVC standard, and after that we propose two kinds of multiple description video coding schemes based on H.264/AVC standard, Specifically including the following two parts of work:1. The MSVC-RP MDC scheme was in-depth studied, and we propose a novel Multiple Description Video Coding scheme with Significance-based Macro-block Classification MDVC-SMC, which improved the redundant coding efficiency of MSVC-RP. Through the importance based macro-block classification method combined with hierarchical QP allocation algorithm, we improved the coding efficiency of the redundant pictures in the MSVC-RP MDC scheme, as a result the redundant coding efficiency of the scheme is also improved. The simulation experiments in the ideal MD network show that, compared with MSVC-RP the proposed MDC scheme achieves better side rate-distortion performance over almost the entire bit-rate range.2. Two other novel region of interest (ROI) coding based multiple description video coding schemes are proposed. Firstly we study the error concealment mechanism of the H.264/AVC standard, and the actual performance of the error concealment method in the restoring of the image outside the ROI is analyzed. Based on the above analysis the common ROI detect algorithms were in-depth studied, and we proposed a region of interesting fast detect algorithm for H.264/AVC standard. Using this ROI fast detect method combined with the temporal information splitting techniques and special information splitting techniques, we design two ROI coding based MDC schemes, and implement the experimental prototype based on JM 17.2. The simulation experiments in the ideal MD network show that, for the video sequences, which have large and still background and there is an obvious region of interest (face), this two schemes can get good rate-distortion performanceThe above multiple description video coding schemes have certain value in the video phone, video conferencing, video surveillance applications and other real-time video applications.