改进的矢量量化码字快速搜索法

Improved fast codeword search algorithm for vector quantization

矢量量化技术是一种高效和有竞争力的数据压缩方法,但由于其编解码过程中需要较大的计算量影响了其使用。提出了一种改进的基于子矢量特征值的码字快速搜索算法。算法充分利用矢量的3个特征值即和值、子矢量和值以及方差,建立起一种5步码字排除法,使得算法能够快速排除大部分不匹配码字,实现减少计算量的目的。仿真实验结果表明,算法的计算量要小于ZhiBin算法、Pan算法以及Chen算法,证明了改进算法的有效性。

Vector Quantization（VQ） is an efficient technique for data compression and retrieval.But its encoding requires expensive computation that greatly limits its practical use.A fast algorithm for VQ encoding on the basis of features of vectors and subvectors is presented.Making use of three characteristics of a vector：the sum,the partial sum and the partial variance,a five-step eliminating algorithm is introduced.The proposed algorithm can reject a lot of codewords.From the experimental results,the proposed algorithm needs less computational complexity against the ZhiBin＇s algorithm,Pan＇s algorithm,Chen＇s algorithm.So the results of the experiment illustrate the rationality and efficiency of the algorithm.