Comments on “Random Distance Method for Generating Unimodal and Multimodal Cancelable Biometric Features”

Abstract

This article points out the fallacies in the theory and its implementation proposed by Kaur and Khanna. They have set out a cancelable biometric-based template protection method to address the security and privacy concerns emerging from the use of biometric systems. There are three major issues associated with the method proposed in their study. The first issue relates to the mathematical fallacy in the proof of the random distance method. The second issue concerns the claim of dimension-reduction by 50%, despite the fact that RDM does not preserve inter- and intra-user variations. The third issue is in salting the feature vectors using the OR operation between the feature vectors and random grid (RG), which is incorrect. As it will result in revealing partial information only and will not increase entropy. Furthermore, they have stated that their approach results in noninvertibility using the median filtering. However, its implementation is flawed.