摘要
矢量量化是一种极其重要的信号压缩方法,广泛应用于图像信号压缩、语音信号压缩等领域,它的主要问题是码本设计问题,传统的LBG算法及其改进算法只能获得局部最优的码本。本文详细讨论了如何利用遗传算法来获得全局最优的码本,目的应用于说话人识别。本文具体给出了基于遗传算法的最优码本设计算法的实现方法和实验结果。实验表明,本文实现的算法有效。
Vector Quantization is a very important signal compression method with rich applications in wide fields such as image signal compression and speech signal compression. The key problem of VQ is the design of codebook. Traditional LBG algorithm yields only local optimal codebook, This paper discusses in detail how to design global optimal codebook making use of Genetic Algorithm for the purpose of speaker recognition, Concrete complementation of the algorithm as well as the experiment is given. Experiment shows that our algorithm implemented is effective.
出处
《信号处理》
CSCD
北大核心
2006年第3期412-416,共5页
Journal of Signal Processing
关键词
矢最量化
遗传算法
最优码本设计
Vector Quantization
Genetic Algorithm
Optimal codebook design
