Enhanced iris recognition method based on multi-unit iris images

Abstract

For the purpose of biometric person identification, iris recognitionuses the unique characteristics of the patterns of the iris; that is,the eye region between the pupil and the sclera. When obtaining an irisimage, the iris's image is frequently rotated because of the user's headroll toward the left or right shoulder. As the rotation of the irisimage leads to circular shifting of the iris features, the accuracy ofiris recognition is degraded. To solve this problem, conventional irisrecognition methods use shifting of the iris feature codes to performthe matching. However, this increases the computational complexity andlevel of false acceptance error. To solve these problems, we propose anovel iris recognition method based on multi-unit iris images. Ourmethod is novel in the following five ways compared with previousmethods. First, to detect both eyes, we use Adaboost and a rapid eyedetector (RED) based on the iris shape feature and integral imaging.Both eyes are detected using RED in the approximate candidate regionthat consists of the binocular region, which is determined by theAdaboost detector. Second, we classify the detected eyes into the leftand right eyes, because the iris patterns in the left and right eyes inthe same person are different, and they are therefore considered asdifferent classes. We can improve the accuracy of iris recognition usingthis pre-classification of the left and right eyes. Third, by measuringthe angle of head roll using the two center positions of the left andright pupils, detected by two circular edge detectors, we obtain theinformation of the iris rotation angle. Fourth, in order to reduce theerror and processing time of iris recognition, adaptive bit-shiftingbased on the measured iris rotation angle is used in feature matching.Fifth, the recognition accuracy is enhanced by the score fusion of theleft and right irises. Experimental results on the iris open database oflow-resolution images showed that the averaged equal error rate of irisrecognition using the proposed method was 4.3006%, which is lower thanthat of other methods. (C) The Authors. Published by SPIE under aCreative Commons Attribution 3.0 Unported License. Distribution orreproduction of this work in whole or in part requires full attributionof the original publication, including its DOI.