Development of Novel Independent Component Analysis Techniques and their Applications

Abstract

Real world problems very often provide minimum information regarding their causes. This is mainly due to the system complexities and noninvasive techniques employed by scientists and engineers to study such systems. Signal and image processing techniques used for analyzing such systems essentially tend to be blind. Earlier, training signal based techniques were used extensively for such analyses. But many times either these training signals are not practicable to be availed by the analyzer or become burden on the system itself. Hence blind signal/image processing techniques are becoming predominant in modern real time systems. In fact, blind signal processing has become a very important topic of research and development in many areas, especially biomedical engineering, medical imaging, speech enhancement, remote sensing, communication systems, exploration seismology, geophysics, econometrics, data mining, sensor networks etc. Blind Signal Processing has three major areas: Blind Signal Separation and Extraction, Independent Component Analysis (ICA) and Multichannel Blind Deconvolution and Equalization. ICA technique has also been typically applied to the other two areas mentioned above. Hence ICA research with its wide range of applications is quite interesting and has been taken up as the central domain of the present work.