Poster: Biometric data transformation for cryptographic domains and its application

Abstract

A large class of biometric template protection algorithms assume that feature vectors are integer valued. However, biometric data is generally represented through real-valued feature vectors. Therefore, secure template constructions are not immediately applicable when feature vectors are composed of real numbers. We propose a generic transformation and extend the domain of biometric template protection algorithms from integer-valued feature vectors to real valued feature vectors. We show that our transformation is accuracy-preserving and verify our theoretical findings by reporting the implementation results using a public keystroke dynamics dataset.