Finger Knuckle Print recognition based on Gabor feature and KPCA+LDA

摘要

Biometric authentication is considered as the most secure and protected method to recognize and verify person's identity. The recent study shows that finger knuckle print of a person can be used as a biometric trait in a biometric authentication system due to its uniqueness property. In this paper, we propose a biometric authentication system which makes use of finger knuckle image of a person as biometric trait. Here feature extraction is done by applying a bank of Gabor filter to a pre-processed FKP image. Then dimensionality of the extracted feature is reduced by using KPCA (kernel principal component analysis) algorithm. Then Linear Descriminant Analysis (LDA) algorithm is applied on KPCA feature space to increase the between class separability features.Euclidean distance measure is used for classification. The proposed system has a recognition rate of 91.67%.