Abstract: A new method for the iris feature extraction and recognition was proposed in this
paper. Firstly, the kernel principal component analysis (KPCA) was used to extract texture
features of iris image because which has the strong ability to extract features in high dimensions
space. In order to reduce the samples of the support venture machine (SVM), the two-layer serial
classifier was designed which combined SVM and distance classification, a rejecting coefficient
and rejecting rule were defined. According to the rejecting rule, the distance classifier could
classify the iris images and give the final results, or reject to classify. The rejected iris images
were fed into SVM for further classification. The classification algorithms could take advantage of
SVM and distance classification. The experimental results show that the method can improve the
rate of iris recognition, and is effective.

Key words: iris recognition, feature extraction, kernel principal component analysis, support
vector machine