Research On Side Information In Distributed Video Coding

Distributed video coding(DVC) is a new video coding paradigm with simpleencoder, high compression efficiency and good error resilience ability. It meets thedemand of some particular applications such as wireless multimedia networks andwireless video surveillance, which have the restriction of capacity and poor calculationof video terminal. Now DVC has become a hot subject of research.In DVC system, side information is of great importance which essentiallyinfluences the system’s rate distortion(RD) performance in several aspects. The moreaccurate the side information is, the fewer parity bits need to be sent to decoder,resulting in better compression performance. Thus how to estimate the side informationaccurately and effectively is a key to DVC. In this paper, amounts of side informationestimation and refinement algorithms are analyzed. To the problem of high complexityin side information refinement(SIR), this paper proposes an improved SIR algorithm fora transform domain WZ video codec where the SI is successively improved along thedecoding of the various DCT coefficient bands without adding significant complexity.The already decoded coefficients provide more accurate information since they arecloser to the original WZ frame than the initially estimated SI. To perform thisrefinement for the remaining DCT coefficients, the proposed technique makes use of thealready decoded coefficients to find out the low accuracy areas in the initial SI, andrefine them with the adjacent key frames. The refined SI is used to decode the next DCTcoefficient band to improve the RD performance. The results show significant andconsistent performance improvements especially under critical conditions such as highmotion content.