多描述编码与单描述编码加FEC的性能比较-信号与信息处理专业论文.docxVIP

控制用的数据（即纠检错码），以实现自动纠错或检错的目的。由此， 我们考虑到多描述编码在信源端就引入冗余是否值得的问题。 本实验在上述理论背景基础上搭建了两个系统，一为传统单描述 编码系统，即 JPEG 信源编码加上 Turbo 信道码；二为多描述编码系 统，编码方法主要是基于 JPEG 编码的多描述变换编码算法。通过系 统一得到了编码原始图像源所需压缩码率大小、信道码 Turbo 为纠错 所引入的冗余度和最后重建图像的 PSNR；通过系统二得到了多描述 编码与传统 JPEG 编码相比所引入的冗余度以及重建图像的 PSNR。 通过设定相同信道丢包环境，在本实验中可以得到以下结论，在引入 冗余度值相差不大的前提下，多描述编码系统的重建图像质量要比单 描述编码好的多；此外，当信道丢包率很小甚至趋于零时，单描述编 码在比多描述编码引入较小冗余度的情况下还能获得更好的重建图 像质量。 考虑到 Turbo 码和既定多描述编码本身的一些缺陷，本文还在上 述两个系统中提出了自己的优化算法以改善系统性能。对于系统一， 我们主要加入了交织解交织模块来“打散”信道丢包的出错位置，充 分发挥 Turbo 码的纠错性能，所重建图像质量在此基础上进一步提 高。对于系统二，我们另外实现了 DC 复制法、子块分类图像估计法； 并设计了 S-Type 配对法、Zigzag 序号配对法。上述方法都从一定程 度上较好地提高了重建图像的质量。 关键词：多描述编码、单描述编码、多描述变换编码 THE PERFORMANCE COMPARISON OF MULTIPLE DESCRIPTION CODING AND SINGLE DESCRIPTION CODING COMBINED WITH FEC Abstract The rapid development of Internet has made image processing one of the mainstream business over network, which thus powerfully promotes the standardization of various coding/decoding technologies. As we all know, the objective of traditional coding/decoding technologies is to improve the reconstructed image quality and to transmit more information under the given channel capacity. With the development of network, the uncertainty in its capacity and packet losing all make the coding technology not only concern the data adaptation to the randomness of network, but give full consideration to overcome the data error during the transmission. Under this situation, overall research has been conducted in scalar coding technology which integrates unequal error protective system. However, since all coding data is divided into basic level and enhanced level, the decoding of enhanced-level data is dependent on its basic level. In any case, if we lose basic-level information, a terrible reconstructed image will be gotten from the decoder. Considering this serious defect in scalar coding, multiple description coding has become a research hotspot since decades ago. It divides single source into multiple independent bit streams, which we call de