A kurtosis-dependent parameterized blind source separation algorithm and stability analysis

In the framework of natural gradient, a KurtosisDependent Parameterized Blind Source Separation (KDPBSS) algorithm is proposed, which can separate the mixture of super- and sub-gaussian sources. A unifying weighed double model is proposed. According to kurtosis value of source and whitening, model parameters are adaptively calculated which can be used to estimate super- and sub-gaussian source distributions and its corresponding score functions directly. According to stability analysis, the ranges of model parameters are fixed which confirm KDPBSS algorithm stable. Applied to the mixture of five images, the experiment shows the proposed algorithm has better performance and convergence than some proposed algorithms.