Introduction to blind source separation
By Juan Manuel Miguel Jimenez, Luciano Boquete Vazquez
The problem of blind source separation consists in obtaining each of the signals generated by n statistically independent sources, from the linear mixtures that take place in the very medium in which they propagate and that are detected at least by n sensors. Various algorithms developed in recent years make this technique a promising tool in a myriad of applications, among which noise filtering, audio signal processing, astrophysical signals, fetal electrocardiography, electroencephalography and the analysis of and enhancement of medical images.