Efficient Algorithm of Eye Image Check for Robust Iris Recognition System

Abstract

For the improvement of iris recognition system performance, the filtering algorithm that picks out counterfeit and noisy data is very important. In this paper, as a part of preprocessing step, we propose the efficient algorithm of eye image check, composed of two stages for detecting the fake and noisy eye data. The first stage is to detect the fake iris data evaluating the coefficient of variation of pupil radius and eyelid movement, and analyzing of 2D Fast Fourier Transform (2D-FFT) spectrum. The second stage is to find out the noisy image such as blink, eyelash interference and the truncation of iris region on the eye image. Using this algorithm, the improvement of about 2% at the accuracy rate of the system is achieved. For the experiment, we integrate the algorithm with iris recognition system, made use of Daubechies’ Wavelet and Support Vector Machines (SVM) for feature extraction and pattern matching. Experiment results involve 1694 eye images of 111 different people and the accuracy rate of 99.1%.