A Novel Iris Recognition System Based on Central Force Optimization

Abstract

Iris Recognition has emerged as a reliable biometric technique in the recent past. Huge amount of work has been carried out continuously to improve its performance and accuracy. This paper proposes an iris recognition system using FFBNN-CFO. CFO is a novel metaheuristic technique which works on the basis of physical kinematics, and can be used for optimization of Neural Network parameters. Initially the eye images are preprocessed using adaptive median filter to remove salt and pepper noise. Then, the statistical features are extracted from the iris and pupil region. These features are then used to train the FFBNN. During training, FFBNN parameters are optimized by CFO to improve the accuracy of the system. In the testing phase the performance of the proposed iris recognition system is tested with the help of sufficient number of eye images from 2 major databases MMU1 and UBIRIS. The performance of the proposed method is compared and contrasted with other methods such as Adaptive Acceleration Particle Swarm Optimization (AAPSO), Particle Swarm Optimization (PSO) and FFBNN. The comparison shows that, the proposed iris recognition system based on FFBNN-CFO, gives higher recognition accuracy as compared to FFBNN-AAPSO, FFBNN-PSO and FFBNN.