IRIS RECOGNITION SYSTEM USING COMBINED HISTOGRAM STATISTICS

Abstract

This paper proposes a novel iris recognition system based on matching colour pixel statistics. The proposed system uses colour histograms as pixel statistic feature vectors for recognition of irises by cross correlation between the histogram of a given iris and the histograms of irises in the database. Majority voting (MV) has been applied to achieve the final recognition. Iris images taken from the UPOL are pre-processed and segmented. The proposed system gives a 100% recognition rate on 64 iris images. The reliability of the proposed system is demonstrated by providing FRR and FAR curves with 0.00% EER.