An Automated Finger-Knuckle-Print Identification System Using Jointly RBF & RFT Classifiers

Ensuring the security of a wide variety of applications such as public security, access control and banking, is becoming an increasingly important problem as modern technology is integrated into the majority of these applications. Thus, establishing identity of a person is the significant method used to ensure a high security level. As a security method, biometric identification has a long tradition and is a synonym for the uniqueness of person. In this paper, based on Finger-Knuckle-Print (FKP), we present a multimodal personal identification system using two feature extraction methods with their fusion applied at the matching score level. In this study, the segmented FKP is firstly represented by the Histogram of Oriented Gradients(HOG) descriptors. Subsequently, the Radial Basis Function(RBF) and Random Forest Transform (RFT) models are used to design two sub-systems. The proposed method is validated for their efficacy on the available PolyU FKP Database of 165 users. Our experimental results show the effectiveness and reliability of the proposed approach, which brings both high identification and accuracy rate.