New Biometric modAlities for Secure Contactless Border Security Clearance

BASIS?s goal is to provide a contactless, fast and seamless framework to capture the desired biometrics in order to tackle both authentication as well as identification of bona-fide travellers. BASIS proposes a screening system where travellers will funnel through a controlled corridor equipped with multiple calibrated & synchronized RGB+D sensors, as well as microphones. As the traveller walks through the sensors, the acquisition system will capture and fuse the data into an accurate 3D representation of the traveller?s head that will provide accurate 3D shape features. These 3D features combined with standard 2D features acquired from the same sensors will be fused into a feature vector. At the same time, while walking inside this corridor the traveller will be required to pronounce his or her name, allowing for an extra voice feature vector. Combining different biometric traits into a combined feature vector, the system will identify or authenticate the traveller. By fusing multiple biometrics, in this case hybrid 2D/3D face, hybrid 2D/3D ear and voice combined with other soft biometrics high verification rates are ensured. In parallel, BASIS will focus on offering a reliable system, trying to minimize malfunctions and maximize passenger throughput, by focusing on the identification/authentication computational needs, and specifically for the identification case effort will be also devoted on the indexing of the database. BASIS will focus on these areas: - Creating a large dataset of calibrated and synchronized multi-sensor captures - On the fly 3D Face Reconstruction - Generic Multi-modal Biometric Fusion in two levels - Contact-less authentication/identification in a two-phase corridor - First phase initial decision - Second phase with additional input needed when initial decision is inconclusive - An adaptive classifier using soft biometrics - Creating two separate prototypes (enrollment + recognition)