An efficient method for feature extraction of human iris patterns

Abstract

A system that automatically recognizes individuals based on biometric traits has been an attractive goal for researchers for a long time. Iris recognition is a biometric identification method that combines computer vision and pattern recognition. It produces one of the most accurate methods available for security systems because of the uniqueness of the human iris. The process of iris recognition is split into 4 major steps. These steps are: Iris segmentation, normalization, feature extraction, and matching. This paper focuses on the step of feature extraction and encoding. A new method is proposed to extract features from the iris image. The method uses a sliding window technique and mathematical operations on the pixels to produce a feature vector. Experimental results of the method produced a relatively small feature vector of size 5×120, which contributes to the efficiency and speed of an iris recognition system, as well as reducing the amount of memory needed. The algorithm written for the method also includes a step to eliminate the effect of varying light intensity, which improves the accuracy of the overall system as well as reduces the time needed to acquire an image with suitable lighting. Other techniques to unify the level of light intensity among all images were applied as well. Evaluation of the method was done by considering various performance metrics such as the false acceptance rate (FAR), false rejection rate (FRR), and the recognition rate of the algorithm. The recognition rate achieved from the proposed method was about 98.54%.