分布式视频编码系统中边信息技术研究-电子与通信工程专业论文.docx

万方数据 万方数据 摘 要 分布式视频编码具有编码端复杂度低、功耗小以及抗误码能力强等特点，相 比于传统的视频编码技术更适用于移动视频通信、视频共享等应用场合，具有广 阔的应用前景，近年来受到了广泛关注。 边信息估计是分布式视频编码中的关键技术之一，估计出的边信息质量直接 影响着整个分布式视频编码系统的率失真性能。当视频序列运动剧烈时，采用运 动补偿内插法估计出的边信息质量较低。解码过程中，解码端获取了更多与原始 WZ 帧相关的信息，因此可以利用这些信息对边信息进行更新。本文以变换域的 Wyner-Ziv 视频编码系统为基础框架，在对已有的边信息估计和更新算法进行研究 的基础上，采用了改进的运动补偿内插法估计边信息，在每完成一个系数子带的 解码后采用改进的边信息迭代更新算法对边信息进行更新，利用相邻关键帧对初 始边信息进行修正，再利用更新后的边信息解码下一个系数子带。仿真结果显示 采用边信息改进算法后边信息的 PSNR 以及系统的率失真性能均得到了提升。 关键词：分布式视频编码 变换域 运动补偿内插 边信息更新 万方数据 万方数据 Abstract Compared with the traditional video coding methods, distributed video coding (DVC) is more suitable for energy constrained applications such as the mobile video communications and video sharing, with the characteristics of low encoding complexity, low power consumption and strong error resilience. DVC has been widespread concerned in recent years and also has a wider application prospect. The side information estimation is one of the key technologies in DVC system, and the quality of side information directly influences the rate distortion performance of a DVC system. The quality of side information generated by motion compensated interpolation is not always good especially when deal with high motion content. In the process of decoding, the decoder obtains more information about the original WZ frame which can be used to refine the initial side information. Based on the transform domain Wyner-Ziv video coding system, this paper has a deep research on the existed algorithms to estimate and refine the side information first, and then uses the improved motion compensated interpolation method to estimate the side information, refine the side information with adjacent key frames through the improved side information refinement algorithm after decoding a coefficient band. At last the refined side information will be used to decode the next coefficient band. The simulation results show that the improved side information algorithms can improve the PSNR of the initial side information and the DVC rate