Securing Fingerprint Template: Fuzzy Vault with Helper Data

An important issue gaining attention in biometrics community is the security and privacy of biometric systems: How robust are these systems against attacks? What happens if the biometric template is lost or stolen? Can the privacy of the users be preserved even when a security breach occurs? Among the numerous attacks that can be launched against these systems, protecting the user template that is stored either locally (e.g., on a smart card) or centrally (e.g., on the server) is a major concern. As a possible solution to this problem, a new class of algorithms, termed biometric cryptosystems has been proposed. These systems do not store the original template but only a transformed version of the template within a cryptographic framework. An example of such systems is the fuzzy vault construct proposed by Juels and Sudan. In this construct, the biometric template is converted to a 2D point cloud, containing a secret such as a symmetric encryption key. The operation of the vault requires some "helper" data. In this paper, we present an implementation of the fuzzy fingerprint vault based on orientation field based helper data that is automatically extracted from the fingerprints. We further show that this helper data does not leak any information about fingerprint minutiae, hence complementing the increased user privacy afforded by the fuzzy fingerprint vault. We demonstrate the vault performance on a public domain fingerprint database.