一种基于CAVLC解码的快速码表查找算法

在分析和研究基于上下文的自适应可变长度编码(CAVLC)码表结构特点的基础上，结合码表统计规律，提出一种新的CAVLC解码码表查找算法。根据码字前缀0的个数和码字长度之间的关系共同决定输入码字后缀位数和数值，实现对输入码流的快速确定，对确定的输入码字，只需再查一次表便可以得到其对应的解码输出。测试结果表明，该优化算法在解码查表速度方面比原算法提高约20%。

A Quick Code Table Lookup Algorithm Based on CAVLC Decoding

Based on the thorough analysis and study on the structural characteristics of Context-based Adaptive Variable Length Coding(CAVLC) code tables, this paper proposes a new table lookup algorithm for CAVLC decoding with the statistical law of code tables. The idea of this algorithm is based on the relationship between the numbers of zero in code prefix and the length of code. With it, it proposes a scheme, in which the code suffix and the code value are determined by the relationship between the numbers of zero in code prefix and the length of code, and it can decode the input code quickly. As a result, a specific input code can be decoded with only one time table lookup. Test results show that the optimized algorithm can achieve 20% speed-up in the table lookup compared with the original algorithm.