多视点视频编码中容错控制技术研究

项目名称： 多视点视频编码中容错控制技术研究

项目编号： No.61301106

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 无线电电子学、电信技术

项目作者： 项欣光

作者单位： 南京理工大学

项目金额： 28万元

中文摘要： 多视点视频可以提供视频场景多角度，全方位的立体视觉，目前已经广泛应用于视频监控，视频会议，自由视点电视和三维电视等众多视频通讯应用中。但是传输网络的不可靠性向多视点视频传输提出了巨大的挑战，容错鲁棒的多视点视频传输成为了多视点视频编码中急需解决的问题。本项目将研究多视点视频编码中容错控制技术，实现多视点视频容错鲁棒性传输。具体研究内容涉及：针对多视点视频的高视间域相关性，设计面向局部视频信息丢失的多视点视频错误隐藏方案，实现对多视点视频局部丢失信息的重构；结合多视点视频编码结构，设计信息矢量推导依赖的多视点视频整帧错误隐藏方案，实现对多视点视频整帧丢失信息的重构；结合率失真和最优化理论，建立多视点视频编码端到解码端失真估计模型，设计多视点视频容错率失真优化及码率控制算法，实现编码性能与容错鲁棒性传输的平衡。本项目可取得理论创新与技术突破，推动多视点视频的发展与应用。

中文关键词： 多视点视频编码；容错控制；错误隐藏；整帧错误隐藏；容错率失真优化

英文摘要： Multi-view video has been used widely in numerous video communication applications, such as surveillance, video conference, free-viewpoint television and 3-D television, by providing a multi-angle video scene and comprehensive stereo vision. However, the unreliability caused by the channel errors of transmission networks has posed a huge challenge for multi-view video transmission. Robust multi-view video transmission over error-prone networks has become an urgent problem need to be resolved in multi-view video coding. This project plans to study error control technologies for multi-view video coding to realize robust multi-view video transmission over error-prone networks. Firstly, we plan to propose error concealment schemes which utilize the high inter-view correlations between different views for multi-view video coding to reconstruct local lost video contents. Then, utilizing the special structure of multi-view video coding, the information derivation based full frame loss error concealment schemes are proposed for multi-view video coding to reconstruct lost entire video frame. Finally, utilizing the rate-distortion and optimization theory, we plan to establish encoder-to-decoder distortion model and design rate-distortion optimized error resilience algorithm and bit-rate control algorithm for multi-view vi

英文关键词： multi-view video coding；error control；error concealment；full frame loss error concealment；rate-distortion optimized error resilience