A blind source separation method based on generalized eigendecomposition in the wavelet domain

In this paper, a blind source separation method based on generalized eigendecomposition in the wavelet domain is proposed. It uses two random numbers as the filter coefficients to realize the linear transform, while the time domain algorithm needs selecting different filter for different source signals. The proposed method is steady, very fast and easy to implement. The computer simulations results demonstrate the correctness and effectiveness of the proposed algorithm.