A fuzzy vault implementation for securing revocable iris templates

Abstract

In recent years biometric cryptosystem evolved as a means for solving key management issues as well as protecting biometric templates. The fuzzy vault is a well known cryptotographic construction well suited for biometric systems. It has been studied theoretically as well as practically implemented in biometric systems using different biometric traits. When implemented in iris-based biometric system, the fuzzy vault faces two main challenges: 1) it requires an unordered set for successful implementation, and 2) it needs to deal with intra-class variations. In this paper, we implement a fuzzy vault based on the iris templates. A modified fuzzy vault resolves the issue of unordered set and error correction coding is used to deal with intra-class variations. An iris shuffling algorithm is also integrated into the system to ensure revocability. The proposed structure can be integrated with existing databases using binary iris templates and hence does not require the redesign of the biometric authentication system. Revocability ensures that even if the system is compromised new templates can be issued without compromising privacy of the individuals. The system is evaluated using the CASIA database and results show that the system is successful in ensuring security and revocability of the iris templates without compromising the performance