SIRIS - Secure IRIS authentication system

Abstract

As the password based authentication systems are not able to meet the performance because they can be stolen, forgotten, cracked, sniffed and tampered with. Lateral thinking to this problem evolved the use of biometrics to authenticate the person uniquely. Since, the templates are stored in a centralized database there is possibility of tampering of templates. The objective of this paper is to combine cryptography with biometrics by apply minor changes to the iris templates to transform them and store in a database, and hence forth even the system is compromised, the template is safe from the wrong hands thus to improve the security of the system in a network. For determining the performance of the system digitized grayscale eye images from CASIA 1.0[3] iris image set is used. This authentication system consists of an automatic recognition of iris template using the password provided by the user. The system performance is tested using different key sizes 128 bit, 256 bit and 512 bit keys. We used Hamming distance for classification of iris templates, and templates are accepted to match if the statistical independence was failed. Experimental results showed that the system performed with perfect recognition on a set of 756 images and resulted in an accuracy rate of 96%. Copyright � 2012 ACM.