Iris segmentation and normalization methods for biometric identification

Abstract

This report is devoted to the analysis of the iris segmentation and normalization methods for application in biometrics identification. As performance of iris recognition systems highly depends on segmentation and normalization, effective feature extraction methods are of great importance for massive databases as CASIA, to increase the overall accuracy. The proposed methods are based were based on assumed pupil non-circularity and application of developed image warping methods. The automatic segmentation algorithm was proposed and developed to localize the iris region and isolate the eyelid and eyelash through the fit-ellipse method and radius-finding method.