Blind source separation of independent/dependent signals using a measure on copulas

We introduce a new BSS approach, based on modified Kullback-Leibler divergence between copula densities, for both independent or dependent source component signals. In the standard case of independent source components, the proposed method improves the mutual information (between probability densities) procedure, and it has the advantage to be naturally generalized to separate mixtures of dependent source components. Simulation results are presented showing the convergence and the efficiency of the proposed algorithms.