The Iris Recognition Technology And Its Application To Identity

The performance of single face recognition is very poor when there is a pose variation. To tackle the problem, the recognition rate improved by using series feature fusion method. Be different from conventional Independent component analysis (ICA1), a method is obtain for statistical independent feature vectors directly (ICA2). ICA1mainly represent part of information of image, while ICA2mainly represent overall information of image, so these two features have quality of complementary. Firstly, two types of complimentary features are extracted using ICA. Then those features under different weighting are concatenated to form a high dimensional fused feature. Finally, finish the identification using the European minimum distance classifier. The good results of ICA experiments on the USTB database4show that face recognition by series feature fusion method turning left15degree and30degree is much higher than corresponding single feature recognition rateIn this dissertation, we discuss the series fusion methods in face based recognition technology before. The main contribution of this dissertation includes:1) Present the situation of the research at present and the summaries the face recognition approaches and the crucial problems in the research.2) Sort out the basic principle of ICA, to establish its mathematical model, analysis experimental concrete steps. Firstly, image pre-processing remove the elimination of light, noise and so on; secondly, data center and whitening remove the observation of high-order correlation signal to obtain a better algorithm convergence; thirdly, introduce FastICA algorithm in detail, and extract the vector-based face image by it; finally, complete face recognition by minimum distance classifier.3) Using USTB image database the face features of ICA1and ICA2in its sub-base is extracted, and reconstruct the corresponding vector-based image in Matlab environment. Complete the face recognition based on ICA1, ICA2and series feature fusion when face turn left15degree and30degree. Analysis the effect of research and bring forth prospect of future work...