等和值块扩展最近邻矢量量化码字搜索算法

论文提出一种等和值块扩展最近邻矢量量化码字搜索算法。该算法将码书按和值大小排序分块,并将每一块中间或中间附近的码字的和值作为本码书块的特征和值。编码时,查找与输入矢量和值距离最近的码书块并作为初始匹配码书块。然后在该码书块附近上下扩展搜索相邻码书块中距输入矢量最近的码字。该算法具有无复杂运算的特点,易于VLSI技术实现。仿真结果表明,该算法是一种有效的码字搜索算法。

Equal-Sum Block-Extending Nearest Neighbor Search Algorithm

A fast vector quantization encoding algorithm,equal-sum block-extending nearest neighbor search algorithm, is presented in this paper.Before the search process,all codewords in the codebook are sorted in the ascending order of their sum,and the codebook is divided into blocks in average.The sum of the mid codeword in each block is regarded as eigenvalue of this block.During the search process,the authors firstly search the block whose eigenvalue is nearest to the sum of the input vector,and initialize the current block to be the closest one.And secondly,we can find out a nearest codeword to the input vector using up-down search mechanism in the blocks near the initial best-matching block.This algorithm without complex operations is easy to be implemented using VLSI technology.Experimental results demonstrate that the proposed algorithm is effective.