IRIS Recognition Based on Kernels of Support Vector Machine

Ensuring security biometrically is essential in most of the authentication and identification scenario. Recognition based on iris patterns is a thrust area of research cause to provide reliable, simple and rapid identification system. Machine learning classification algorithm of support vector machine [SVM] is applied in this work for personal identification. The profuse as well as unique patterns of iris are acquired and stored in the form of matrix template which contains 4800 elements for each iris. The row vectors of 2400 elements are passed as inputs to SVM classifier. The SVM generates separate classes for each user and performs matching based on the template's unique spectral features of iris. The experimental results of this proposed work illustrate a better performance of 98.5% compared to the existing methods such as hamming distance, local binary pattern and various kernels of SVM. The popular CASIA (Chinese Academy of Sciences - Institute of Automation) iris database with fifty users' eye image samples are experimented to prove, that the least Square method of Quadratic kernel based SVM is comparatively better with minimal true rejection rate.