Research On Channel Model And Side Information Improvement For Distributed Video Coding

Distributed video coding (DVC) is a new mode of video encoding paradigm based on two major information theory results:the distributed lossless coding of Slepian-Wolf and lossy coding with decoder side information of Wyner-Ziv Theorems. It adopts intra-frame coding and inter-frame coding, which shifts the computational burden from the encoder to the decoder, and it takes advantage in low encoder complexity, low power consumption and strong error resilience. Thus, DVC has its application in wireless video terminator such as wireless video monitoring system and video sensor network which have limited computation ability, storage capability and power consumption, and also has developed rapidly in recent years with the growth of these new video applications.This dissertation focuses on the modeling of virtual channel and improvement of side information based on the research of principle and framework for distributed video coding, and the main research and contributions are listed as follows.1. The modeling of virtual channel of DVC has been studied. Based on the theory of Gaussian channel model and Laplacian channel model, I comprehensive studied the method of that how to estimate the channel parameters for Laplacian channel with both off-line (encoder-generated) and on-line (decoder-generated) techniques, which work at different granularity levels. And a new channel modeling algorithm used in transform domain is proposed.2. The generation and optimization programs for side information have been studied, and an improved side information refinement using color-texture features for DVC was presented to improve the quality of the initial side information. The content analysis technique is used to segment the initial SI into different regions and further to find the inaccurate motion vectors, then a modified 3-D recursive search block matching algorithm is implemented to improve the motion vectors. With more reliable motion vectors, the adaptive OBMC approach is proposed to enhance the performance of side-information generation algorithm. Finally, the quality was verified by simulation through different video sequences.3. A DVC system is implemented based on the proposed virtual channel and side information refinement. LDPC code is applied on Wyner-Ziv codec in a transform domain, and the principle of each module is presented. A matlab system simulation has been done to the DVC system, by comparing with typical DVC scheme and the intra-frame coding of H.264 standard and the inter-frame coding of H.264 standard, proves the validity and feasibility of the proposed DVC system.