Research Of Correlation Noise Estimation For Distributed Video Coding

Distributed video coding(DVC)is rapidly gaining popularity as a low cost,robust video coding solution.DVC is built on distributed source coding(DSC)principles where correlation between sources to be compressed is exploited at the decoder side.In the case of DVC,a current frame available only at the encoder is estimated at the decoder with side information generated from other frames available at the decoder.However,the state-of-the-art WZ code designs perform well only when correlation statistics between sources are stationary and known at the encoder and decoder.we have a problem of WZ coding with nonstationary correlation noise with unknown statistics since the scene dynamically and unpredictably changes.Therefore,how to accurately predict and track correlated parameters is very important for DVC systems.This dissertation conducts researchs on the distributed video coding technology and correlation estimation methods,proposing a adaptive correlation estimation with sliding-window for DVC,which can predicting and tracking correlation noise parameter and improve the iterative decoding process and the system rate-distortion performance.The specific research contents and work are as follows:(1)Considering that correlation noise modeling between source and side information is non-stationary in distributed video coding,the on-the-fly correlation estimation method is introduced into the Slepian-Wolf decoding process,thus this paper makes research on the non-stationary correlation estimation algorithm.Different from the previous offline estimation and static online estimation,on-the-fly estimation uses the current bits to estimate correlation parameters in the decoding process.In turn,the optimized parameters speed up iterative convergence.In the asymmetric Slepian-Wolf coding based on LDPCA code,this paper focuses on an advanced on-the-fly correlation estimation algorithms,named sliding-window belief propagation algorithm,and improves the objective function and search strategy of the optimal window selection of it.Finally,comparison and analysis of performances are performed through experiments.(2)Integrating sliding-window with distributed video coding,this dissertation propose an adaptive correlation estimation algorithm with sliding-window in DVC.This algorithm aims at the non-stationarity of DVC system and is embedded in the Slepian-Wolf decoder itself.In the decoding process,the statistical parameters of the correlated noise are estimated on-linewith accompanying substreams;as the iterative decoding process continues,correlation estimation will become more accurate,in turn,accurate correlation parameters can improve the accuracy of decoding;the method of sliding-window can accelerate the calculation,and the setting ofthreshold also can adaptively select reestimate,which saves unnecessary expenses,thereby optimizing the iterative decoding process and improving the performance of DVC system.(3)Embedthe proposed algorithm into the DVC architecture and implement it.In this system,the encoder uses the LDPCA code to encode and control the bit-plane bit rate,and the decoder uses the adjusted joint bit-plane belief propagation algorithm for iterative decoding.I demonstrate the proposed scheme within state-of-the-art DVC systems,which are transform domain based with a feedback channel for rate adaptation.Experimental results show that our proposed scheme outperforms the benchmark state-of-the-art DISCOVER codec and other cases without correlation tracking in terms of the overall rate distortion performance,estimation accuracy,and decoding efficiency,achieving comparable decoding performance but with significantly low complexity comparing with sampling method.