ABSTRACT

In different fields, the problem of Blind Source Separation (BSS) is known as a collection of linear combinations of unknown sources and worse the coefficients of the linear combinations that are unknown. The main problem is to estimate the matrix of the combination coefficients (mixing matrix) and to reconstruct the sources according to it. The quality of separation of sources from mixtures dramatically is done by exploiting the quality of sparsity of sources, whereby the sources are properly represented according to some collection of signals. There are two main approaches to the “blind source separation” problem solution they are clustering and ICA (independent component analysis).