A fast algorithm for blind extraction of smooth complex sources with application in EEG conditioning

A fast complex-valued blind source extraction algorithm designed for the extraction of smooth sources, is introduced. This is achieved based on a smoothness constrained cost and by considering the augmented statistics of the latent source signals. The methodology, based on the CR calculus and complex FastICA framework, is shown to be capable of extraction of both complex circular and noncircular smooth sources for real-time brain computer interfacing. The performance of the algorithm is verified on benchmark signals, followed by a biomedical application in the extraction of artifacts from recorded EEG signals in the order of smoothness.