Combined Noise Reduction and Joint Diagonalization Techniques for Blind Source Separation

Source separation from Instantaneous combinations of them, so called observations, is a challenging problem in signal processing. In many cases there is not information about the mixing matrix and observations must process blindly, so in these cases problem is called Blind Source Separation (BSS). Another most important problem in signal processing is existence of noise, so by now the elimination and reduction of noise techniques have been introduced too much. In this paper we try to combine Blind Source Separation, especially joint diagonalization technique, and wavelet based noise reduction technique, for better source separation. Simulation results show effectiveness of proposed method.