Skip to main content

MultiBiometric Signal Exploitation, Analysis and Acceleration

Researcher: Marios Savvides

Research Area: Security of Cyber-Physical Systems

Abstract

Scope: This project aims to explore several areas in Multi-Biometrics to determine reliable biometric signals for enhancing cyber-physical security, force protection and vulnerability analysis. More importantly this project aims to be a starter for multiple seed thrusts for developing a proof of concept prototypes (and demos) which each will be used to establish larger federal grants for completion. One of the new thrusts (not explored to-date) includes utilizing Graphical Processing Power (in high end Graphics cards and professional nVidia TESLA cards) which can provide cheap, cost-effective, supercomputing power on a single desktop with the aim of providing up to 2 orders of magnitude in speed up of current Biometric technology. This will allow us to also attempt problems that were up to now thought to be too complex to in achieve real-time. Overall, we have research planned in areas of (1) Continuous User Authentication, (2) GPU Accelerated Biometric Algorithms (face detection, recognition and iris), (3) Object/Person/Car Tracking, (4) Periocular Recognition, (5) MultiFactor Tensor & Manifold Learning, (6) Off Angle Facial Landmarking, Cellphone based face detection & recognition, (7) Optimal Feature Extraction Optimization, (8) Iris Spoofing, (9) Feasibility Analysis for Long Range Iris Acquisition (20m-50m and beyond).

Outcomes: Proof of concept demonstrations that can be used to show to external sponsors and obtain followon funding for full-fledged prototypes. Conference/Journal Papers and book chapters.