Confronting a Variety of Finger Vein Recognition Algorithms With Wax Presentation Attack Artefacts

Abstract

Presentation attacks for finger vein recognition systems has emerged to be a widely researched topic. For this reason, a previously published, shown to be non-functional, approach for presentation attack artefact generation has been redesigned and used as a basis to generate a new publicly available finger vein presentation attack database. In order to assess the threat emitted by those artefacts from a broad perspective, an extensive vulnerability analysis including twelve finger vein recognition algorithms, that can be grouped into three meta categories of algorithms, was conducted. Experiments on the group of vein structure-based recognition algorithms, that also includes the de facto standard feature extraction method for vulnerability analysis, indicate a high level of threat by Impostor Attack Presentation Match Rates up to 90%. However, fundamentally different approaches (i.e. more relying on the entire finger texture under NIR illumination) show little to no susceptibility at all. This indicates that the actual threat level represented by presentation attack artefacts has to be re-considered in the light of these results.