An Iris Authentication Scheme Based on Fuzzy Extractor

Fuzzy extractors allow one to extract some uniformly distributed random key in an error-tolerant way from a biometric input w and then successfully reproduce the key from any other biometric input w′ that is close to w. In this paper, a fuzzy extractor based iris authentication scheme is provided, which combines error correcting codes and cryptography, when Hamming distance is adopted as the biometric matching metric. The impacts of iris intra-class differences upon false rejection rate are analyzed and a two-layer coding scheme in which iterative codes and Reed-Solomon codes are applied is presented. In place of introducing mask filters to reduce iris code bit error rates, error correction capability is enhanced to strive for more iris valid entropy, when quality of iris image is not very good because of disturbance of eyelashes, eyelids, etc. The scheme is evaluated using 4096-bit iris codes from 128 different eyes, with 3 samples from each eye. Simulation experiments show that the false rejection rate is less than one percent. Additionally, both security and privacy of user’s biometric template can be well protected, and user registration update can also be supported. The proposed scheme is especially applicable to iris authentication system with privacy requirements.