An iris recognition algorithm using local extreme points

Abstract

The performance of an iris recognition algorithm depends greatly on its classification ability as well as speed. In this paper, an iris recognition algorithm using local extreme points is proposed. It first detects the local extreme points along the angular direction as key points. Then, the sample vector along the angular direction is encoded into a binary feature vector according to the surface trend (gradient) characterized by the local extreme points. Finally, the Hamming distance between two iris patterns is calculated to make a decision. Extensive experimental results show the high performance of the proposed method in terms of accuracy and speed. © Springer-Verlag Berlin Heidelberg 2004.