The study and test of ICA algorithms
Hong Li, Yunlian Sun · 2005
Independent component analysis (ICA) is a recently developed method used to separate unknown source signals from their mixtures. This paper describes some representational algorithms of ICA in brief, such as the fast-ICA, the EGLD-ICA, the kernel-ICA and the MS-ICA. The simulative tests show that the application of ICA in the fields of signal processing and communication, such as extraction weak signal, separation in audio signals and digital image watermarking, is widely and effectively. At the same time, the performances of these kinds of ICA algorithm are contrasted and the results are shown by the form of figures and tables in this paper. And the conclusion tries to discuss the selection of type of ICA algorithm in different practical application.