Fuzzy vault and template-level fusion applied to a binary fingerprint representation

Abstract

The fuzzy vault is an error-tolerant authentication scheme that stores data in a privacy-preserving way. It allows to hide and retrieve a secret by presenting a set of elements that can slightly differ. This paper aims at applying the scheme to a transformation-invariant binary fingerprint representation. We also show how to perform multi-finger fusion with no extra space requirement and perform fuzzy vault in this case. Our results are quite encouraging with respect to the inherent constraint of decreasing False Accept attacks on template protection schemes in order to ensure higher privacy.