基于自适应冗余数量和参数估计的容错编码  

Error resilience coding based on adaptive redundancy allocation and parameter estimation

作　　者：徐巍炜[1,2] 田翔[1,2] 陈耀武[1,2] 

机构地区：[1]浙江大学数字技术及仪器研究所,浙江杭州310027 [2]浙江大学浙江省网络多媒体技术研究重点实验室,浙江杭州310027

出　　处：《华中科技大学学报（自然科学版）》2014年第6期68-72,共5页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金仪器专项资助项目(40927001);浙江省重点科技创新团队资助项目(2011R09021-06);中央高校基本科研业务费专项资金资助项目(2011FZA4002)

摘　　要：针对视频流在通过易错信道传输时丢包和延时导致的降低接收端的视频质量问题,为了提升视频流的容错性能,提出了一种快速容错编码策略,对每个主要帧自适应地编码多个冗余帧,并且进一步提出了快速多冗余帧数量计算的方法.利用实验数据得到失真和码流消耗的函数,使用编码参数来快速估计率失真代价值,进而根据失真模型对具有不同冗余帧计数的冗余帧的编码参数进行限制.由于序列相邻帧的特征具有相似性,因此使用临近帧的编码参数及失真和码流消耗对统计模型进行更新.最终通过利用失真和码流消耗的统计模型,分析冗余帧数量对总体端对端率失真代价的影响,以此选择优化的冗余帧数目和编码参数.实验结果表明:该算法与自适应调整编码参数的冗余帧编码的算法相比,峰值信噪比平均提高约1dB,具有更好的端对端视频传输质量.Quality degradation occurs during transmission of video streaming over the error-prone network.A fast error resilience coding method was proposed to solve this,different amount of redundant pictures were adaptively adopted,and the optimized method was proposed to decline computational cost on calculating redundant picture number.By analyzing the feature of distortion and bit consumption,the statistical module of rate-distortion cost was formulated as the function of encoding parameters,and then the encoding parameters of redundant pictures with different redundant picture counts were limited based on the model of rate-distortion cost.The encoding parameters,distortion and bit consumption of neighbor frames were used for the updating of statistical model.The optimized number of redundant picture was finally chosen by estimating the influence of redundant picture number on the total rate-distortion cost.From experimental results,the average PSNR（power signal-to-noise ratio）of the proposed method outperforms the state-of-the-art redundant coding method by about 1dB over error prone channel.

关 键 词：视频编码 容错编码 端对端失真 冗余片 冗余帧 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]