Optimization and evaluation of a biometric recognition technique based on 3D Ultrasound Palm Vein

Biometric systems based on ultrasound images are becoming more and more common and are exploiting different characteristics like fingerprint, palmprint, hand geometry and palm vein patterns. In this work, a previously proposed method for extracting 3D palm vein patterns from volumetric ultrasound images has been optimized and experimentally evaluated by using a home-made database composed on 55 volumetric images captured from 23 different volunteers. If compared with recent works, recognition results seems encouraging. Further possible improvements of the proposed technique are finally discussed.