多描述分布式视频编码分析-通信与信息系统专业论文.docxVIP

摘 要 多描述视频编码是一种有效的传输差错控制技术，它可以在无线信道中实现 鲁棒性传输。多描述编码将信息源分成多路描述，每个描述通过相互独立的信道 进行传输且都有恢复整幅图像的能力，因此，即使在传输过程中丢失了某一路描 述，仍然可以通过其他的描述来恢复图像，而多个描述结合则可得到更高质量的 重建视频信号。 当前多描述视频编码方案大都基于多个预测环路，这些方案在理想多描述信 道中表现良好。然而，在无线传输中，由于信道丢包而导致解码端和编码端的参 考帧不同，会造成多描述编码端和解码端的预测误匹配，这个偏差随着时间扩大， 会对视频图像重建造成严重的影响。本文利用分布式视频编码来代替传统的预测 编码，提出了一种无反馈多描述分布式编码系统，不仅克服了预测误匹配的问题， 而且采用基于感兴趣区域的多描述分配算法和循环移位处理得到了两路平衡的描 述，解决了在多描述分布式编码中因为分布式视频编码引入的反馈信道问题；同 时，针对如何协调描述间独立性和相关性的问题，进一步提出一种可自动调节冗 余的算法，该算法通过简单运动估计模块得到帧的运动量来动态调整相关因子， 实现描述之间冗余的自动调节，使系统性能达到最优。仿真表明，算法在信道丢 包率高的情况下，保证了视频信号的传输质量，去除反馈信道后整体性能下降较 少，保证了鲁棒性传输和实时传输。 关键词：多描述编码 预测误匹配 分布式视频编码 鲁棒性传输 实时传输 Abstract In wireless video transmission, multiple description coding (MDC) is an effective way to enhance the error resilience over packet loss channels. Multiple description coding divides the information source into several descriptions, each of which is transmitted through an independent channel and have the ability to restore the whole image. Therefore, even if some descriptions are lost in the transmission, an acceptable quality can be achieved through other descriptions. While all the descriptions are received, a better quality can be yielded. Now multiple description video coding schemes are mostly based on multiple prediction loops and behave well in the ideal multiple description channel. However, In the wireless transmission, the channel packet loss will lead to difference between the reference frame in the encoder and the decoder, which will cause a drift problem. The drift propagates with time, and may cause a serious damage of the whole video sequence. Using distributed video coding to replace the traditional predictive coding, this paper proposed a novel feedback free multiple description distributed video coding to beat the drift problem, in which a cycle shift algorithm and extracting the regions of interest in bitplanes are used to produce balanced descriptions and remove the feedback channel in DVC scheme. Meanwhile a simple video motion estimation method i