Generalized contrast functions for blind separation of overdetermined linear mixtures with unknown numbers of sources
M. Sahmoudi · IEEE/SP 13th Workshop on Statistical Signal Processing, 2005 · 2005
In this paper, we address the problem of blind separation of m independent sources from their n linear mixtures in the overdetermined systems (n ges m) with unknown number of sources. After generalizing the definition of classical and nonsymmetrical contrast functions, we exhibit a wide class of generalized contrast functions using some superadditive functional and concave functions. Two practical generalized contrasts based on the support Lebesgue measure (SLM) and the mutual information (MI) criteria are proposed and discussed. Finally, computer simulations illustrate the results and demonstrate all the interest we can find in considering a generalized contrast function