一种可实现零内存存取的CAVLC解码算法  

A CAVLC Decoding Algorithm with Zero Memory Access

作　　者：黄明政[1] 韩一石[1] 

机构地区：[1]广东工业大学信息工程学院,广州510006

出　　处：《计算机工程》2014年第3期278-282,共5页Computer Engineering

基　　金：广东省科技计划基金资助项目(2011B090400344;2011B010200029)

摘　　要：在基于上下文的自适应可变长度编码(CAVLC)解码算法中,对非结构化自适应可变长度编码码表进行解码时需要反复查找码表进行码字匹配,从而导致解码速度慢和需要大量内存存取的问题。为此,提出一种可实现零内存存取的CAVLC解码算法。将CAVLC码字前缀0的个数作为一级索引,同时通过一级索引获得输入码流的可能长度。将码字后缀作为二级索引并获得码字的值,直接通过码字快速获得解码结果。对于确定的输入码字,只需通过无码表查找代码操作即可得到对应的解码输出。测试结果表明,该算法不仅可以实现零内存存取的CAVLC解码,而且其解码速度比标准算法提高了45%。This paper proposes a zero memory access algorithm for direct Context-based Adaptive Variable Length Coding（CAVLC） decoding, aiming to solve the problems of slow decoding speed and a large of memory accesses caused by the repeated table look-up for the match codeword during the unstructured variable length coding table decoding. This algorithm takes the numbers of zero in CAVLC code prefix as the primary index, and gets the probably length of the input code stream. It takes the code suffix as the secondary index and gets the codeword value. It can get the decoded result by the codeword quickly. For the specific input code, the decoded output can be directly gotten without any table look-up. Test results show that not only the algorithm can achieve zero memory access for CAVLC decoding, but also the decoding speed of this algorithm can achieve 45% speed-up, compared with the standard algorithm.

关 键 词：基于上下文的自适应可变长度编码 零内存存取 码字前缀 一级索引 码字后缀 二级索引 

分 类 号：TP919.81[自动化与计算机技术]