A KPCA-Based Biometric Identifier Using Finger Knuckle Surface

Biometrics deals with human intrinsic physical or behavioral patterns. It is divided into physiological and behavioral types. Biometric authentication using finger knuckle surface is a new technique. The fingers have to be kept wide open while imaging. The image has to be taken in a white background. The knuckle surface of each human being is unique in nature. Various finger geometrical calculations are needed for the extraction of knuckles. In this approach various techniques used for feature extraction of finger knuckle surface is mentioned in a detailed manner. The various techniques used here are ICA (Independent Component Analysis), LDA (Linear Discriminant Analysis), PCA (Principal Component Analysis) and KPCA (Kernel Principal Component Analysis). The main aim is to find which feature extraction technique gives better performance in case of finger knuckle surface authentication. The usage of KPCA method helps in attaining better results. The experimental results graphs are plotted for each knuckle separately and which technique gives better performance is noted.