Face Recognition Based On ICA

Traditional man face recognition technology with complex algorithm is not precise and robust. In order to improve the technology, the paper describes the man face recognition technology bases on blind signal process. This technology is based on the ICA and KICA.ICA recognition technology abstracts the independent character as characteristic vectors. ICA algorithm with excellent partial character expression ability can make use of the hidden information between pixels. The fixed point ICA algorithm the paper refers to is based on fixed point calculation method. The existence of fixed point ICA algorithm makes the analysis of high dimensionality data possible. This algorithm has some feature, such as: good astringency and less computation.KICA recognition method is based on the nonlinear function space. It is not only kernel ICA but also a whole new ICA method. KICA implement the nonlinear transform with kernel functions meet the kernel qualification. KICA recognition technology also has good astringency and very high success ratio of recognition.The chapter 3.4 shows the simulation data of algorithm this paper refers to.