An Unsupervised Iris Based Biometric System with Inherent Feature Thresholding

Abstract

This paper provides a general indication of the existing approaches rely on basic factor (i.e. extraction of iris information, affine transform, and distance matrix) as input. An essential factor in effective design of IRIS based biometric approach is the accuracy with which the model can estimate a region of interest (i.e. IRIS) within constraints and unforeseen issues, which can be very problematical but need of the hour. We introduced a new IRIS based biometric system that incorporates various factors that takes complete information of eye for developing the feature set (digest) while affine transforms are not incorporated while the three sets distance measure is incepted to enhance accuracy. The algorithm-based size of template, functionality of distance measure and/or scope, methods and/or function of application through well-defined scientific and statistical principles. Unfortunately, the accuracy of the existing approaches is limited despite the large scale of experience with several improvements based on digital image processing and statistical models. Henceforth, we incorporated several texture analysis algorithms with computing techniques along with several parametric enhancement constraints to ensure the feasibility, effectiveness and efficiency of the proposed framework in comparison with the existing methods.