Methods for Source Number Estimation in Underdetermined Blind Sources Separation

Estimate the number of source signals is a necessary prerequisite for underdetermined blind sources separation (UBSS). The accuracy of sources number estimation has influence to the correctness of the sources separation. For this, a new algorithm—Hough-Windowed is proposed based on the assumption that the source signals are sparse. First, the algorithm constructs straight line equations of the observed signals based on Hough transformation. In order to obtain cluster areas, histogram is cumulated by windowed in transform domain. Then estimate the maximum of every cluster area. The number of different maximum is the number of source signals. Simulation results show the validity and expansibility of the algorithm. At the same time, compared with Potential function, the algorithm reflects the better noise immunity and the lower sparse sensitivity.

[1]  Barak A. Pearlmutter,et al.  Hard-LOST: modified k-means for oriented lines , 2004 .

[2]  Elena Cordero,et al.  Sparsity of Gabor representation of Schrodinger propagators , 2009 .

[3]  Yuzo Hirai,et al.  Blind source separation by a geometrical method , 2002, Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290).

[4]  Xie Shengli,et al.  Sparse representation and blind source separation of ill-posed mixtures , 2006 .

[5]  Tao Xiong,et al.  A combined SVM and LDA approach for classification , 2005, Proceedings. 2005 IEEE International Joint Conference on Neural Networks, 2005..

[6]  Michael Zibulevsky,et al.  Underdetermined blind source separation using sparse representations , 2001, Signal Process..

[7]  Aapo Hyvärinen,et al.  Fast and robust fixed-point algorithms for independent component analysis , 1999, IEEE Trans. Neural Networks.

[8]  Erkki Oja,et al.  Independent component analysis: algorithms and applications , 2000, Neural Networks.