ACCURACY OF IRIS RECOGNITION USING TRAPEZOIDAL TEMPLATES

Abstract

Authentication is the process of identifying someone or something and verifying its validity.There are a lot biometric technologies, which include finger scanning, finger vein ID, facial recognition, voice recognition, retina scanning, iris recognition etc. Being one of the most reliable methods, we have chosen to analyze iris recognition. Since it is unique, unchangeable and difficult to falsify, it is considered to be a great candidate for authentication. This process involves several steps – segmentation, normalization, encoding, matching. The quality of iris pictures noninheritable at-a-distance or below less strained imaging environments is understood to degrade the iris recognition accuracy. The periocular data is inherently embedded in such iris pictures and might be exploited to help within the iris recognition below such non-ideal situations. Our main aims in this thesis are new methods for the segmentation and encoding stages of the iris recognition using trapezoidal templates. Considering all the newly presented methods, we can say that they improve the performance of the algorithm for the CASIA database.