矢量量化快速搜索算法的研究

Fast search algorithm for vector quantization

为了进一步降低矢量量化的运算量,提出了一种新的快速搜索算法。在最近邻域搜索算法的基础上,提出了一个基于矢量分割的一般性码字排除准则。该准则综合利用子矢量的均值和方差参数,构造了一个判决不等式来排除不可能的码字。算法中子矢量的个数设定为2。实验结果表明,该算法的运算时间是改进的等均值等方差最近邻域搜索(IEENNS)算法的80%左右。该算法的性能要优于以往的几种基于不等式判决的快速搜索算法,可以应用在语音和图像编码算法中。

A fast search algorithm was developed to reduce the computing load for vector quantization. The algorithm uses a general codeword rejection theorembased on the nearest-neighbor search algorithm and vector division. The theoremuses the means and variances of sub-vectors to form an inequality to reject unlikely codewords. The fast search algorithm is designed when the number of subvectors has been reduced to two. The experimental results show that the computing time for the new algorithm is 80% that of the improved equal-average equal-variance nearest-neighbor search (IEENNS) algorithm. The new algorithm has higher performance than other fast algorithms based on the results of inequalities and can be used in speech and image coding.