基于对象的H.264视频编码抗分组丢失方法  

Object based H.264 video coding for packet loss resilience

作　　者：艾达[1] 常义林[1] 罗忠[2] 王静[2] 

机构地区：[1]西安电子科技大学综合业务网理论与关键技术国家重点实验室,陕西西安710071 [2]华为科技有限公司,广东深圳518129

出　　处：《西安电子科技大学学报》2006年第4期524-528,共5页Journal of Xidian University

基　　金：国家自然科学基金资助项目(61072030);华为科技基金资助项目

摘　　要：为了改善视频在IP网络传输时分组丢失造成的质量下降,提出了一种计算复杂度低且简单易行的视频对象划分新方法,以宏块编码类型作为视频对象划分的依据,将图像划分为前景和背景两个视频对象.结合H.264新的抗误码技术———灵活的宏块次序,将前景和背景对象分别置于不同的Slice组中,使视频对象可独立解码并采用不同的抗分组丢失方案.在此基础上提出了基于视频对象的信源-信道联合编码方法,加强对前景对象的强制Intra刷新率,并根据信道丢包率采用不同码率的信道编码对视频对象分别进行保护,有效地提高了分组丢失网络中的视频通信质量.A novel method for video object segmentation is proposed for the purpose of packet loss resilience, which partitions a frame into foreground and background objects according to their Macroblock encoding mode. Combined with the new H. 264 error resilience technique Flexible Macroblock Ordering, the two objects are put into different Slice groups that enable the independent packet loss resilience method for them. In the proposed joint source-channel coding method a more forced Intra update rate is applied on the foreground and different coding rates of channel encoding are used to the two objects with the packet loss rate. The technique presented in this paper is easily implemented and it improves the quality of video communications in the packet loss IP network effectively.

关 键 词：视频对象 视频通信 分组丢失 H.264 灵活的宏块次序 

分 类 号：TN919.3[电子电信—通信与信息系统]