An Effective Feature Extraction Approach for Iris Recognition System

Background/Objectives: Iris recognition system refers to a system used for identifying different iris texture patterns for different applications. The research is aimed at developing a system that improves the efficiency of the iris recognition system and make it more reliable and robust. Methods: To this end, we have developed a system based on compound local binary pattern technique. Compound local binary technique is a spatial domain technique, which assign a 2P bit code to central pixel based on the local neighbourhood comprising of P neighbours. The operator takes into consideration both the sign and magnitude information of the central and corresponding neighbour grey values. The unique and abundant features extracted through Compound Local Binary Pattern (CLBP) operator act as input to the neural network classifier. Findings: The system has been tested over 50 eye images taken from CASIA database. Iris recognition system based on Compound local binary pattern technique along with neural network used as classifier improves the accuracy of system in comparison to existing feature extraction approaches. In this proposed research, recognition rate achieved is 96%.