Automatic Feature Reduction Framework For Identification Process In Palm Vein Recognition

Abstract

Abstract

Feature or dimensionality reduction has become one of fundamental problem in the field of pattern recogni- tion such as biometrics. The choosing number of fea- ture or dimension has become one challenge. Instead choosing number of feature manually, in this paper, we proposed an automatic feature reduction by using a cas- caded feature reduction schemes based on variance or- der of the DCT feature space and eigenvalue of k-PCA in the palm vein recognition. Based on experiment re- sults, our proposed scheme can achieve recognition rate above 0.92 accuracy which uses fewer features and can reduce time process significantly until 99.5% comparing with traditional manual feature reduction method.