Blind source separation algorithm based on information canonical correlation analysis

被引：0
作者：
Bai, Zhimao [1 ,3 ]
Huang, Gaoming [2 ]
Xu, Qinzhen [3 ]
Yang, Lüxi [3 ]
机构：
[1] School of Public Health, Southeast University
[2] College of Electronic Engineering, Naval University of Engineering
[3] School of Information Science and Engineering, Southeast University
来源：
Dongnan Daxue Xuebao (Ziran Kexue Ban)/Journal of Southeast University (Natural Science Edition) | 2009年 / 39卷 / 06期
关键词：
Blind source separation; Information canonical correlation analysis; Mutual information;
D O I：
10.3969/j.issn.1001-0505.2009.06.002
中图分类号：
学科分类号：
摘要：
To solve the problem of blind source separation, a novel algorithm based on information canonical correlation analysis (ICCA) is presented by combining the theory of mutual information with canonical correlation analysis. In this algorithm, the information canonical vectors are searched out by maximizing the mutual information between the linear combination of the observed vectors and the linear combination of the delayed observed vectors. The probability density function is estimated by Gaussian kernel estimates. Then, the source signals are extracted and separated one by one by multiplying the information canonical vectors with the observed mixture. The simulation results show that this algorithm can separate the mixture signals which consist of super-Gaussian components or of sub-Gaussian components. The mixture signals including these two components and the ill mixture signals can also be separated effectively.
引用
收藏
页码：1093 / 1097
页数：4
相关论文
共 14 条
  • [1] Jutten C., Herault J., Blind separation of sources. Part I: An adaptive algorithm based on neuromimetic architecture, Signal Processing, 24, 1, pp. 1-10, (1991)
  • [2] Comon P., Independent component analysis - A new concept, Signal Processing, 36, 3, pp. 287-314, (1994)
  • [3] Hyvarinen A., Gaussian moments for noisy independent component analysis, IEEE Letters on Signal Processing, 6, 6, pp. 145-147, (1999)
  • [4] Even J., Saruwatari H., Shikano K., Blind signal extraction via direct mutual information minimization, IEEE Workshop on Machine Learning for Signal Processing, pp. 49-54, (2008)
  • [5] Chien J.T., Hsien H.L., Furui S., A new mutual information measure for independent component analysis, IEEE International Conference on Acoustics, Speech and Signal Processing, pp. 1817-1820, (2008)
  • [6] Duarte L.T., Jutten C., A mutual information minimization approach for a class of nonlinear recurrent separating systems, IEEE Workshop on Machine Learning for Signal Processing, pp. 122-127, (2007)
  • [7] Huang G., Yang L., He Z., Time difference direction finding based on canonical correlation analysis, Journal of Circuits and Systems, 10, 5, pp. 10-15, (2005)
  • [8] Liu W., Mandic D.P., Cichocki A., Analysis and online realization of the CCA approach for blind source separation, IEEE Transactions on Neural Networks, 18, 5, pp. 1505-1510, (2007)
  • [9] Via J., Santamaria I., Perez J., Deterministic CCA-based algorithms for blind equalization of FIR-MIMO channels, IEEE Transactions on Signal Processing, 55, 7, pp. 3867-3878, (2007)
  • [10] Li Y.O., Wang W., Adali T., Et al., CCA for joint blind source separation of multiple datasets with application to group FMRI analysis, IEEE International Conference on Acoustics, Speech and Signal Processing, pp. 1837-1840, (2008)
12