Moments-Based Feature Vector Extraction for Iris Recognition

Abstract

Biometric recognition is a personal identification system which serves as a prime authentication method for a number of applications. Finding a unique biometric trait that can support classification across a large dataset is always a problem in biometric recognition system. Iris is one such biometric trait that is unique over a large dataset. Mathematical moments are used to extract features from the iris region surrounding the pupil. These moments help to capture a large information on the distribution of texture on the iris region. Based on this moment features we perform iris recognition using nearest neighbour classifier. This proposed method with hard threshold achieves an overall recognition rate 84% of and shows scope for improvement.