Proof of concept of 3D ultrasensitive Doppler imaging of fingertip vasculature for biometric identification

Fingerprint-based biometric systems are widely used for personal recognition. However, increasing concerns have been raised regarding their susceptibility to spoofing. In an era where digital security is crucial, this calls for new techniques to be evaluated. A number of studies have focused on vascular biometrics instead, as it is invisible to the eye and therefore difficult to reproduce. Most of this research has focused on palm vein or wrist vein vasculature captured using near infrared spectroscopy, which is limited to 2D projections, and prone to blurring and shading artifacts. In this work we propose a novel methodology based on ultrasensitive Doppler acquisitions, imaging the finger vasculature in 3D. We evaluated the feasibility of using such data to build a biometric authentication system based on finger vasculature mapping, with a Siamese neural network for biometric matching. Our results show that such a system can reach high matching accuracies (Equal Error Rate of 1.28%) without the need for image registration or reproducible finger positioning.