首页 | 本学科首页   官方微博 | 高级检索  
     

基于熵编码CABAC的信源信道联合解码器
引用本文:王粤,解蓉. 基于熵编码CABAC的信源信道联合解码器[J]. 上海交通大学学报, 2013, 47(7): 1143-1148
作者姓名:王粤  解蓉
作者单位:(1.浙江工商大学 信息与电子工程学院,杭州 310018; 2.上海交通大学 电子信息与电气工程学院,上海 200240)
基金项目:国家自然科学基金资助项目
摘    要:H.264的熵编码都采用基于上下文自适应二进制算术编码(CABAC),能达到较高的压缩性能,但对信道误码非常敏感.文中提出了一种基于CABAC的算数码变长码联合解码算法,联合信源信道算数码解码之后的信息作为变长码的输入信息,再通过变长码格状图搜索获得最佳的符号序列.同时,在算数码解码部分可以利用变长码的码字结构信息来删除无效搜索路径,提高解码性能.仿真实验表明,该联合迭代解码算法明显优于传统的分离解码器.

关 键 词:联合信源信道解码   上下文自适应二进制算术编码   变长码   高斯白噪声信道  
收稿时间:2012-05-11

Joint Source Channel Decoder Based on CABAC
WANG Yue,XIE Rong. Joint Source Channel Decoder Based on CABAC[J]. Journal of Shanghai Jiaotong University, 2013, 47(7): 1143-1148
Authors:WANG Yue  XIE Rong
Affiliation: (1. School of Information & Electronic Engineering, Zhejiang Gongshang University, Hangzhou 300018, China;2. School of Electronic Information and Electric Engineering, Shanghai Jiaotong University, Shanghai 200240, China)
Abstract:H.264 has been widely used in recently years. The entropy coding method used in H.264 is context based adaptive binary arithmetic code (CABAC). Although CABAC can achieve high compression, it is very sensitive to channel errors. In this paper, a novel joint arithmetic and variable length decoding algorithm based on CABAC was proposed. The transmitted sequence bits were first decoded by a joint source channel arithmetic decoder, the output of which was decoded by joint variable length decoder (JVLD). The trellis graph was used to search the best symbol sequence. Furthermore, at the process of joint arithmetic decoding, the searching path was canceled if the decoded sequence did not conform to the vlc structure, thus achieving better decoding performance. Experimental result indicates that the proposed algorithm is superior to the separate scheme.
Keywords:joint source channel decoding(JSCD)  context-based adaptive binary arithmetic coding (CABAC)  variable length code(VLC)  additive white Gaussian noise(AWGN)
本文献已被 CNKI 万方数据 等数据库收录!
点击此处可从《上海交通大学学报》浏览原始摘要信息
点击此处可从《上海交通大学学报》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号