Cancelable biometrics for finger vein recognition

Abstract

Cancelable biometrics is one of the possible solutions to security and privacy problems in biometrics-based recognition systems. In this paper we propose the use of two classical transformations, block re-mapping and image warping, for the definition of cancelable biometrics from finger vein pattern images. Specifically, the impact on matching performance of the employed distortions, as well as the effects of their parameters' selection, are here investigated. An analysis of the renewability of the employed approaches is also provided. Performance comparable with what achieved in the unprotected approach can be reached in case of block re-mapping transformation, which is also able to provide renewability.