In this paper we introduce a novel cancellable biometric realisation approach based on the iterated inner products between the tokenised pseudo-random number and the face feature to produce a set of user-specific compact binary code, coined as FaceHash. This approach enables straightforward revocation of FaceHash via token replacement and there is no deterministic way to reveal FaceHash without having both tokenised random number and face feature, thus offers strong protection against biometrics fabrication. In addition, FaceHashing has significant functional advantages over contemporary biometrics like zero error rate and clean separation of the genuine and imposter populations.
All Science Journal Classification (ASJC) codes
- Signal Processing
- Computer Vision and Pattern Recognition
- Artificial Intelligence