Iris Segmentation

Abstract

AbstractTwo main challenges of iris segmentation of realistic eye images are processing speed and segmentation accuracy which have been addressed through two methods of iris segmentation in this chapter. The first method (using Canny edge detector) is primarily aiming at faster iris segmentation of more realistic images of UBIRIS database with sufficient segmentation accuracy. Second method is based on the pupil dynamics which has been used for fake iris detection but not for iris segmentation. This method has provided an accurate and loss-free iris segmentation. Unlike most of the other methods, this method does not assume circular or elliptical shape of iris. The performance of this method is tested on UPOL database images, which have high resolution and very low intensity gradient at iris boundaries. This novel method is also capable of fake iris detection.KeywordsIris segmentationOuter boundary localizationPupil detectionSpecular reflection removalIris normalizationCanny edge detectionPupil dynamicsCustomized iris segmentationAccurate iris segmentation using pupil dynamicsCASIA databaseUPOL databaseUBIRIS databaseFake iris detection