【作者】 刘达富；

【导师】 彭强；

【作者基本信息】 西南交通大学 ， 交通信息工程及控制， 2003， 硕士

【摘要】 采用运动补偿预测及可变长编码的压缩视频数据，在极易发生干扰的网络环境中传输时，信道传输差错对压缩编码图像质量影响很大，有时甚至会导致整个视频通信完全失效。解决这一问题的视频差错恢复和掩盖技术因而正成为当前视频压缩及通信领域中十分关键、活跃的前沿技术。本论文以新一代视频压缩标准JVT(Joint Video Team)测试模型JM1.4为基础，重点研究了视频差错恢复和掩盖技术，其中包括解码器端差错掩盖技术和基于编码器／解码器交互的无等待ARQ技术。 本文内容安排如下： 第一章简要介绍了视频压缩带来的问题及相应的解决办法； 第二章和第三章分别详细介绍了JVT视频编码标准中引进的新算法及目前视频差错恢复和掩盖技术的内容、方法和发展现状。这些知识是进行本论文研究的基础； 第四章重点研究解码器端差错掩盖技术，并针对目前差错掩盖技术本身存在的一些不足，引入了边框匹配SM、预处理PREP和运动向量外插MVE，提出了搜索SEARCH等相应的改进算法；然后以JM1.4为实验平台，实现并测试了这些改进算法。实验结果表明，各改进算法能不同程度地提高和改善重建图像的峰值信噪比(PSNR)和主观质量； 第五章研究、实现并测试了基于JVT视频编码标准的无等待ARQ技术。测试结果表明：本文实现的无等待ARQ算法相对于传统的ARQ技术和差错掩盖技术而言，能在不影响解码器端重建图像正常回放的情况下，阻止差错的扩散，实现图像的完全彻底恢复。更多还原

【Abstract】 Loss of information during transmission of compressed video over error prone networks may lead to objectionable visual distortion in the decoded video, and such distortion can propagate to many frames because of the motion compensated prediction and the variable-length coding used. As the solution to this serious problem, video error recovery and concealment techniques thus play a key role and attract more and more attentions in current video compression and communication. In this paper, research on such video error recovery and concealment techniques as decoder-side error concealment (EC) and feedback-based retransmission without waiting (NWARQ) is carried out based on the test modal reference software JM1.4 of the developing ISO/IEC MPEG and ITU-T VCEG video coding standard JVT (Joint Video Team).In chapter 1 a brief introduction to problems resulting from video compression and transmission and solutions to the stated problems is addressed. Detailed descriptions of JM1.4’s newly adopted algorithms are presented in chapter 2. Chapter 3 discusses various video error recovery and concealment techniques, including error resilient encoding schemes, receiver-side post-processing and several feedback-based error control techniques, etc. Knowledge in these chapters forms the theoretical basis of our research.In chapter 4 the emphasis of our research is mainly concentrated on EC technique. Several theoretically useful improvements over current EC techniques’ deficiencies are introduced, such as Side Matching, Preprocessing and Motion Vector Extrapolation, and Search at Decoder Side is proposed. All methods presented in this chapter are implemented and tested based on modified JM1.4 codec of JVT. Simulation results show that both objective gains measured in PSNR (peak signal-to-noise ratio) and visual improvements of reconstructed video can be achieved by utilizing introduced methods.Chapter 5 gives the implementation of JVT-based NWARQ technique. Simulation results based on modified JM1.4 codec show that implemented NWARQ can stop error propagation at decoder and completely recover damaged video without introducing any additional playout delay, comparing to EC and conventional ARQ (automatic repeat request) technique.更多还原