Research On Several Key Technologies For Distributed Video Coding

With the development of new technologies such as wireless digital video camera, wireless sensor networks and multiview video streaming, more and more applications require a low-complexity encoder in term of the constraints of computing ability, bandwidth, and power consumption. As the result conventional video coding standards such as MPEG, H.264/AVC become unsuitable for the applications of these applications. In order to reduce the complexity of the encoder, distributed video coding performs intra-frame encoding but inter-frame decoding. Video frames are encoded separately. Computation complexity is thus shifted from the encoder to the decoder since decoded frames are used to generate side information to exploit source statistics at the decoder. Distributed video coding has low encoder complexity and good error resilience but it has a poor coding efficiency and higher decoding complexity than the conventional video coding schemes.. Therefore how to improve the coding efficiency and reduce the decoding complexity of distributed video coding is a main research emphasis of distributed video coding.This thesis focuses on the research of the theoretical basis, key technologies, rate control and reconstructuion algorithms of distributed video coding,.The main contents are below:(1) Introduce the research status of distributed video coding;study the theoretical basis of distributed video coding and introduce some common distributed video coding schemes.(2) Study several key technologies of distributed video coding and propose some problems and something need to be improved.(3) The decoder rate control algorithm can increase the decoding complexity and cause some decoder-to-encoder transmission delay since rate control is performed at the decoder using a feedback channel. This thesis proposes an encoder rate control algorithm for distributed video coding. This algorithm performs rate control at the encoder without use of a feedback channel. The performance obtained by this algorithm is comparable to that obtained using the decoder rate control algorithm at the cost of slightly increased encoding complexity.(4) This thesis proposes a novel reconstruction algorithm for distributed video coding. In this algorithm, pixel values are integers in the interval [0, 255]. We also present a method to model the correlation noise at the encoder to obtain a more accurate estimation of the correlation noise to improve the efficiency of distributed video coding.