Robust Blind Separation of Statistically Dependent Sources using Dual Tree Wavelets

摘要: Blind source separation (BSS) problem is commonly solved by means of independent component analysis (ICA) assuming statistically and non-Gaussian sources. The strict independence assumption can be relaxed to existence subbands where signals are less dependent. In this paper, we use dual tree complex wavelets for the subband decomposition observed small cummulant based approximation mutual information finding most subband(s). We compare proposed method previously reported shift invariant decimated wavelet packet approach, as well innovations approach. found scheme an efficient robust solution BSS dependent One important application related unsupervised segmentation medical remotely sensed multispectral images.