基于前向纠错的多描述编码(FEC-MDC)是一种在包丢失严重的信道中传输可分级图像和视频数据的有效方法.本文针对单一信源多用户的网络应用模型,研究了在描述数N固定的情况下,根据各信道传输码率的不同,通过调整发送包的长度L实现最佳码流传输的问题,提出了一种码流快速优化分割的改进算法.该算法基于已计算出的参考信道码流分割方案,在期望失真最小的准则下,首先通过在各个目标码率的邻域进行搜索计算,将搜索域分割为低码率部分和高码率部分,然后在高码率部分进行粗的二次搜索,得到最终码流分割,这不仅减少了搜索次数,降低了计算的复杂度,而且保证搜索到在该码率情况下的最佳分割.实验结果表明,本文提出的改进算法与之前的方法相比能够得到相同的平均PSNR,但是总的运算时间减少了近40%.%FEC-based multiple description coding ( FEC-MDC) is an effective approach for sending scalable image and video dato over packet-loss networks. For the scenario where different clients access the server via separate links, we envestigate best stream transmission by adjusting packet length with fixed descriptions N, according to the channel rates, and propose a modified fast, nearly optimal stream partition algorithm. Based on the already computed optimal partition of a reference channel, the algorithm calculate the expected distortion in searching area of each target rate, devides the searching areas into low rate part and high rate part, and then, carries coarse secondary search, obtaines best stream partition afterwards. In this way, not only the searching time being saved, computational conmplexity being decreased, the optimal stream partitions are guaranteed for each channel rate. Experimental results show, the proposed algorithm holds comparable performance to known algorithms , while the total computation time are decreased to approximately 40％.
展开▼
