English

应用科学学报 ›› 1999, Vol. 17 ›› Issue (2): 156-162.

• 论文 • 上一篇    下一篇

基于信息理论准则的盲源分离方法

刘琚1,, 鲁子奕1, 何振亚1, 梅良模2   

  1. 1. 东南大学;
    2. 山东大学
  • 收稿日期:1997-11-23 修回日期:1998-02-04 出版日期:1999-06-30 发布日期:1999-06-30
  • 基金资助:
    国家自然科学基金资助项目

An Information Theory Based Method for Blind Separation of Sources

LIU JU1,, LU ZIYI1, HE ZHENYA1, MEI LIANGMO2   

  1. 1. Department of Radio Engineering, Southeast University, Nanjing 210018;
    2. Department of Physics, Shandong University, Jinan 250100
  • Received:1997-11-23 Revised:1998-02-04 Online:1999-06-30 Published:1999-06-30

PDF

91

被引次数

24

摘要： 根据统计独立的假设,利用概率密度函数的Edgeworth展开,得出了基于互信息最小化的目标函数.该函数具有明确的物理意义,它的最小化可以使输出相互独立.仿真实验结果表明所提出的方法可较好地用于盲源分离问题.

关键词: 互信息, 独立信号, 盲源分离

Abstract: Based on the assumption of statistics independent, we proposed an mutual information minimizing object function where Edgeworth expansion is used. This function can be explained clearly and its minimization would make the outputs independent. Computer simulation shows the validity of proposed method.

Key words: mutual information, independent signals, blind separation of sources