A novel algorithm of mixture signal separation based on joint diagonalization

被引:0
|
作者
Liu Ding [1 ]
Zhao Yan [1 ]
Qian Fu-cai [1 ]
机构
[1] Xian Univ Technol, Res Ctr Informat & Control Engn, Xian 710048, Peoples R China
来源
Proceedings of 2004 Chinese Control and Decision Conference | 2004年
关键词
blind signal separation; joint diagonalization; eigenvalue decomposition;
D O I
暂无
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
An algorithm is proposed based on joint diagonalization from the statistic principles, which convert BSS to the problem of the eigenvalue decomposition. Comparing with some other methods, every order cumulant is taken into account in the algorithm. The experiments show that the algorithm is better than other methods in separation effect and its convergence spewed is faster than some others.
引用
收藏
页码:231 / 234
页数:4
相关论文
共 5 条
  • [1] ANDREAS, 2003, ICA2003
  • [2] A class of neural networks for independent component analysis
    Karhunen, J
    Oja, E
    Wang, LY
    Vigario, R
    Joutsensalo, J
    [J]. IEEE TRANSACTIONS ON NEURAL NETWORKS, 1997, 8 (03): : 486 - 504
  • [3] PAJUNEN P, 1999, ICA 99, P267
  • [4] Blind separation of Gaussian sources via second-order statistics with asymptotically optimal weighting
    Yeredor, A
    [J]. IEEE SIGNAL PROCESSING LETTERS, 2000, 7 (07) : 197 - 200
  • [5] ZIEHE A, 1998, ICANN, V98, P675