Contourlet Transform Based Feature Extraction Method for Finger Knuckle Recognition System

Hand based Biometric systems are considered to be more advantageous due to its high accuracy rate and rapidity in recognition. Finger knuckle Print (FKP) is defined as a set of inherent dermal patterns present in the outer surface of the Proximal Inter Phalangeal joint (PIP) of a person's finger back region which serves as a distinctive biometric identifier. This paper contributes a Contourlet Transform based Feature Extraction Method (CTFEM) which initially decomposes the captured finger knuckle print image that results in low and high frequency contourlet coefficients with different scales and various angles are obtained. Secondly, the Principle Component Analysis (PCA) is further used to reduce the dimensionality of the obtained coefficients and finally matching is performed using Euclidean distance. Extensive experiments are carried out using PolyU FKP database and the obtained experimental results confirm that, the proposed CTFEM approach shows an high genuine acceptance rate of 98.72 %.