Dictionary Attack on Functional Transform-Based Cancelable Fingerprint Templates

Abstract

ABSTRACTRecently, Ratha and others proposed a cancelable biometrics scheme which transforms an original fingerprint template into a new one using a noninvertible transformation. However, we show that the original template is recovered by a dictionary attack if two transformed templates originating from it are revealed. In our attack, we simulate the transformation and construct a set of possible pre-images for each transformed template. Then, we find the correct preimage by computing the intersection of these sets. We present an algorithm implementing this idea as well as successful experimental results.