Open Competition 3 - Information Technology
THE NEXT GENERATION BIOMETRIC--Infrared Identification: Accurate, Fast, Scalable, and Secure
Develop a technology for biometric recognition of faces that uses thermal infrared imaging to map unique vascular patterns, operates in real time, identifies anyone whose infrared or visual image is recorded in its database, and is invulnerable to forgery, disguise, and varying environmental conditions.
Sponsor: Infrared Identification, Inc.PO Box 1038
Lorton, VA 22199
A fully automated, reliable, and foolproof capability to identify individuals within various and selected environments is critical to achieving strong national security. Existing methods of identification--primarily biometric, including fingerprinting, hand geometry, iris scanning, and face recognition--currently fail to meet the most stringent security specifications. Infrared Identification (IRID) Inc. proposes to develop a "facial fingerprinting" technology for biometric identification of faces that uses visible and thermal infrared imaging of unique vascular features, operates in real time, and recognizes anyone whose infrared or visual image is recorded in its database. Because the IR image shows any attempt to reconfigure major blood vessels, recognition cannot be evaded or fooled by forgery, or disguise, as can occur using the visible spectrum for facial recognition. Compared to visual face-recognition systems, IRID's vascular infrared imaging will be less vulnerable to varying conditions, such as head angle, expression, or lighting. In accuracy, speed, scalability (number of faces in the database), ease of use, and security against deception, IRID's technology for producing facial biometric templates promises to be a great improvement over current biometrics. Technical risks include finding unique and persistent vascular features that allow accurate matching, producing adequate imaging of vascular features using available infrared cameras, and achieving the computational speed needed to validate subjects in real time against a database of 1 million facial images. IRID will be supported by the University of Virginia, which will focus on the work related to the vascular structure, and West Virginia University, which will support performance verification of the biometric system. IRID does not have the internal resources to completely fund this two-year project, and potential investors find the technical risks to be too high. ATP funding is essential to conducting research of sufficient scope and scale and would accelerate R&D by five or more years. Recognition through patterns made by large blood vessels, the first of three techniques to be developed, would itself match or exceed the reliability of live-scan fingerprinting but has the advantage of not requiring a subject's cooperation. Success using small-vessel patterns or minutiae would produce the first biometric capable of unattended access control. Even with partial success, this project could have dramatic effects on counter-terrorist and crime-fighting activities and provide economic benefits through better national security, reduced identity theft, and improved safety and security at airports and other public buildings.