Design of a New Biometric System for Identity Detection Based on Finger-knuckle-print Using Autocorrelation Features and Local Binary Patterns

Biometric authenticationis an approach for recognizing a person’s identity as the most secure method.One of the newest biometric identifier, which is recently used for personal identity authentication, is finger-knuckle-print.In this paper, we present an efficient method for personal identification which includes autocorrelation features, local binary patterns,combination of kernel principle component analysis and linear discriminant analysis algorithms.Feature extraction is done using zeroth and first-order moments, autocorrelation features, and local binary patterns. Then, dimensionality reduction is done using kernel principle component analysis. The next step, linear discriminant analysis algorithm is applied to increase the separability of features andχ2distance measure is used as a classifier for matching. Poly-U finger-knuckle-print database is used to examine the performance of the proposed method. The result ofexperimentsshows 98.93% detection rate which demonstrate the efficiency of the proposed method in compare to the other approaches.