A Robust K-plane Clustering Algorithm for Blind Separation of Underdetermined Mixtures of Sparse Sources

In this paper, a robust K-plane clustering algorithm has been proposed for blind separation of underdetermined mixtures of sparse sources. In the presence of noise, based on the insufficient sparsity assumption of the source signals, the K-dimensional concentration hyperplanes have been found by using the algorithm, and then using them to estimate the mixing matrix. Simulation results show that the proposed algorithm can provide a good performance for underdetermined blind sources separation when the sources are insufficiently sparse signals.