DCT子域辐射状分割的图像多描述编码算法  

Multiple description image coding algorithm based on radial segmentation in DCT subdomain

作　　者：江玉珍[1] 杨群生[2] 

机构地区：[1]韩山师范学院数学与信息技术系,广东潮州521041 [2]广东工业大学计算机学院,广州510090

出　　处：《计算机工程与应用》2008年第18期166-168,共3页Computer Engineering and Applications

基　　金：国家自然科学基金(the National Natural Science Foundation of China under Grant No.60403045);广东省自然科学基金(the Natural Science Foundation of Guangdong Province of China under Grant No.05006349)

摘　　要：多描述编码(MDC)是有效提高不可靠网络中数据接收质量的一种新方法。结合JPEG编码中DCT子域系数的分布特点,在已有算法(MDBDS)的基础上提出了DCT子域辐射状分割系数的4描述图像MDC算法。该算法采用DC系数重复分配、缺失AC系数补零的方法获取较好的描述重构效果。实验证明,在相同压缩码率下,该算法的整体性能优于MDEZW、MDBDS等同类MDC算法。Multiple Description image Coding （MDC） is a new method which can effectively improve the data receival quality in the irresponsible networks.Combined with the coefficients distribution characteristic of DCT subdomain in JPEG coding and the existing MDBDS algorithm,the paper presents a 4-descriptions image MDC algorithm based on coefficients radial segmentation in DCT subdomain.With the repeating distribution of DC coefficients and putting zeros to the absent AC coefficients,the method gets nice description reconstruction effects.Experiment results prove its general feasibilities are better than the MDEZW and MDBDS algorithms at the same compressed rate.

关 键 词：多描述编码 DCT子域 零叉树 压缩码率 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]