Cancellable Biometrics for Finger Vein Recognition—Application in the Feature Domain

Abstract

Privacy preservation is a key issue that has to be addressed in biometric recognition systems. Template protection schemes are a suitable way to tackle this task. Various template protection approaches originally proposed for other biometric modalities have been adopted to the domain of vascular pattern recognition. Cancellable biometrics are one class of these schemes. In this chapter, several cancellable biometrics methods like block re-mapping and block warping are applied in the feature domain. The results are compared to previous results obtained by the use of the same methods in the image domain regarding recognition performance, unlinkability and the level of privacy protection. The experiments are conducted using several well-established finger vein recognition systems on two publicly available datasets. Furthermore, an analysis regarding subject- versus system-dependent keys in terms of security and recognition performance is done.