基于比特概率优化算法的分布式视频编码

分布式视频编码器中Turbo码纠错所需要的校验位数量直接决定整个编码器的率失真(RD,Rate-Distortion)性能.分析了传统算法的次优化问题,提出一种新的基于比特概率优化的信道似然值计算算法,该算法首先将原始像素进行比特面分解,然后分别对每个比特面进行独立Turbo码编码及联合解码.解码端在已知已解码比特面的条件下,计算当前解码比特面更准确的信道似然值作为Turbo码的解码输入,减少解码所需要传输的校验位数量,提高编码器RD性能.实验结果表明所提算法能明显提升编码系统的RD性能.

Distributed video coding based on bit probability optimization

In distributed video coding scenario,the quantity of parity bits for error correcting directly decides the rate-distortion(RD) performance of the codec.Upon analyzing the sub-optimization of traditional algorithm,this paper has proposed a new algorithm to compute the channel-likelihood of received symbols based on bit probability optimization.Firstly the original pixels are extracted into bitplanes,then each bitplane is turbo encoded independently and decoded jointly.Taking the decoded bitplane as prior inf...