Efficient comparison of encrypted biometric templates

Abstract

Recently, a large amount of security relevant systems assure the permitted access to sensible data by biometric approaches. As also the biometric data itself deserve a high degree of protection, these data are stored encrypted by so-called template protection techniques in the database. But, such an encryption impedes the comparison of two biometric data instances significantly, and therefore we need advanced approaches to apply template protection techniques for identification purposes. In this paper, we present an efficient identification solution that is based on encryped minutiae data of fingerprints, called ECEBT. We evaluate our algorithm on synthetic data concerning multiple noise effects, and on the real world biometric database FVC-2002 DB1 concerning efficiency and effectiveness. © 2012 Springer-Verlag.