Review of ICA Based Fixed-Point Algorithm for Blind Separation of Mixed Images
Chao Ma, Lianmin Wang · International Conference on Bioinformatics and Biomedical Engineering · 2010
The blind separation of mixed images is a very exciting area of research. However, classical techniques such as eigen and singular value decomposition, which are based on second order statistics, fail to blindly separate mixed signals in many circumstances. A rapidly developed statistical method during last few years, Independent Component Analysis (ICA), which is based on higher order statistics, aims at searching for the components in the mixed signals that are statistically as independent from each other as possible. This paper introduces the fundamental theory and basic model of ICA, and analyzes the math principle of frequently-used fast fixed point algorithm for ICA, and applies the algorithm in blind separation of randomly mixed images. The results shows that the algorithm is very effective and reliable.