基于标量量化的多描述视频编码算法  

Multiple Description Video Coding Algorithm Based on Scalar Quantization

作　　者：骆超[1] 史萍[1] 邵丽娜[1] 侯铭睿[1] 

机构地区：[1]中国传媒大学信息工程学院,北京100024

出　　处：《电视技术》2011年第2期33-36,50,共5页Video Engineering

基　　金：亚洲传媒研究中心重点项目(AM0501)

摘　　要：介绍了多描述编码的理论基础,提出了一种基于标量量化的多描述视频编码方法。该方法将待传输视频量化为两个不相同但互有冗余信息的粗量化版本,并分别对这两个版本进行编码和传输,最后利用其冗余信息进行联合解码。仿真结果表明,在高丢包率条件下,提出的多描述视频编码方案的PSNR明显高于传统视频编码。The principles of multiple description coding are introduced in this paper, and multiple description video coding algorithm based on scalar quantization （MDVSQ） is presented. In this method, two rough-quantitative versions of source video stream, which are not totally different, are yielded and transmitted after encoding respectively. Joint decoding is performed by taking advantage of the redundant information of the two versions. The simulation results show that the PSNR of reconstructed image in MDVSQ is obviously higher than traditional video coding at high packet loss rate.

关 键 词：多描述视频编码 标量量化 索引分配 

分 类 号：TN919.8[电子电信—通信与信息系统]