摘要
本文提出了基于改进禁止搜索 (TS)算法的矢量量化 (VQ)码书设计方法 .禁止搜索算法的关键是如何定义一个解以及如何在当前解的基础上生成邻域解 .由于码书设计的两个优化准则是最邻近条件和聚类质心条件 ,本文提出了两种禁止搜索算法的解描述方案 ,其相应算法分别叫基于码书的禁止搜索 (CB TS)算法和基于聚类划分的禁止搜索 (PB TS)算法 .为了提高禁止搜索算法的性能 ,文中在禁止搜索算法中融入了模拟退火 (SA)机制 .为了进一步提高码书性能 ,文中还将码书设计的传统LBG算法融入禁止搜索算法中 .结果表明 ,基于禁止搜索的两种码书设计方案所生成的码书性能都比LBG算法有明显提高 .
Codebook design algorithms based on tabu search (TS) approach are presented for vector quantization (VQ).The key questions of the tabu search are how to define a solution and how to generate neighboring solutions of the current solution.The two optimality criteria of codebook design are nearest neighbor condition and centroid condition,so two corresponding methods to generate new solutions that are called codebook based tabu search (CB TS) algorithm and partitioning based tabu search (PB TS) algorithm are presented respectively.In order to improve the performance of the tabu search algorithm,simulated annealing technique is introduced.In order to improve the codebook performance,the conventional LBG algorithm is introduced in each iteration of the tabu search algorithm.Tests show that the proposed two algorithms based on the tabu search approach can greatly improve the codebook performance compared to the conventional LBG algorithm.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2000年第9期108-110,共3页
Acta Electronica Sinica
关键词
矢量量化
码书设计
禁止搜索算法
vector quantization
codebook design
tabu search
simulated annealing
