Encrypting fingerprint minutiae templates by random quantization

Abstract

An encryption method is proposed to use random quantization to generate diversified and renewable templates from fingerprint minutiae. The method first achieves absolute pre-alignment over local minutiae quadruplets (called minutiae vicinities) in the original template, resulting in a fixed-length feature vector for each vicinity; and second quantizes the feature vector into binary bits by random quantization; and last post-processes the resultant binary vector in a length tunable way to obtain a protected minutia. Experiments on the fingerprint database FVC2002DB2-A demonstrate the desirable biometric performance achieved by the proposed method. © 2010 Springer-Verlag.