Fusion Analysis of a Palmprint-Hand Geometry Multimodal Ultrasound Recognition System

Multimodal biometric systems are employed in several application fields due to series of advantages in terms of universality and recognition rate. Compared to other technologies, Ultrasound has many advantages, because it allows to obtain 3D images of human body and to verify liveness. In this work, a multimodal ultrasound recognition system based on the fusion of palmprint and hand geometry is proposed and experimentally evaluated. Volumetric hand images are acquired by using water as coupling medium and 2D rendering are obtained at various depth levels. Successively, 2D and 3D features are extracted both for palmprint and hand-geometry. Recognition performances are evaluated through verification experiments by employing a homemade database, first by considering the single characteristics and successively by fusing them. Results demonstrated a drastic improving of results with the fusion but not necessarily the best fusion result is obtained by considering the best palmprint and hand geometry.