The Study Of The Side Information In Wyner-Ziv Coding

Emerging applications such as wireless surveillance systems, video sensornetworks and mobile cameras among others, lead to urgent need of the low complexityvideo coding study. Adapting to these requirements, Distributed Video Coding (DVC)attracts many research efforts. DVC is a new coding paradigm based on two majorinformation theory results: the distributed lossless coding of Slepian-Wolf and lossycoding with decoder side information of Wyner-Ziv Theorems. A particular case ofdistributed video coding is the Wyner-Ziv video coding, as lossy coding with decoderside information, making exploiting the redundancy existing in a video sequence at thedecoder comes true. Thus, independent encoding can proceed in encoder, combineddecoding in decoder. So complexity transferring from encoder to decoder being truesatisfiedthenewapplications.This paper aims to research the side information estimation of Wyner-Ziv coding.First the whole developing history of DVC is introduced in this thesis. Then thebackground and basic theories of Wyner-Ziv is studied, especially on the theories ofside information estimation based on hash and frame interpolation, and it has mainlycompleted three aspects of the work as follows: Firstly, a lot of researches have beendone to aim at the side information estimation. Secondly, using the theories of the sideinformation estimation based on hash and frame interpolation, and through many timesof codings and experiments, we have accomplished the corresponding platform, and weevaluate the performance between the two theories. The theory of Frame interpolationhas a good performance, but its computational complexity is high, on the contrary, thetheorybasedonhashisoppositetoit.Finally,throughapixeldomainWyner-Zivcodec,it is verified which is the performance of the side information estimation. In thisexperiment, the side information is estimated bythe method of frame interpolation, andcontrasttotheH.264intraatlast.Fromabove,theimportanceofthesideinformationtotheimagedecodinghasbeenverified,andthisexperimenthasachievedagoodresult.