Iris localization for direction and deformation independence based on polynomial curve fitting and singleton expansion

Abstract

In an authentic biometric system, iris recognition aims to detect the iris pattern of a person. The single unique pattern of the human iris may be extracted from the image and encoded, such that a given code may be compared to several others, and then validating if the patterns belong to a particular eye or not. Iris localization is an important aspect of iris recognition since accuracy in iris localization affects iris recognition. The previous iris localization methods were less efficient owing to slow processing time and inefficiency in handling non-straight faces in imperfect conditions. In this paper, we propose a new iris localization method with direction and deformation independence. It is based on the idea that the iris is localized from the side face and from distance. A novel approach of curve fitting using polynomial along with singleton expansion is adopted to efficiently and accurately localize the iris in any distance and direction from the camera. We validate the method by experimental analysis on the basis of accuracy, segmentation error and execution time. The method is suggested to be significant for diagnosing several eye-related disorders as well as for biometric authentication processes.