Multiple biometric measurements — including facial recognition, voiceprint analysis and even fingerprints or iris scans — are stored in a secured database. When you present your identification card (or a driver's license or a passport) at a checkpoint, the number on that ID is used to call up your biometric templates from the database. The stored measurements are then compared to your live biometric measurements taken at the checkpoint to ensure that you are, in fact, who you claim to be.
It's a positive identification scenario that could become a reality even without having to reissue hundreds of cards or having to adopt new card technologies on which biometric information is stored. Proposing this authentication solution is Sonic Foundry Inc., a Madison, Wis.-based supplier of software tools for creating, editing and processing digital multimedia. The company's main business involves working with movie, broadcast and sound studios, where digital technologies are making content available to today's rapidly-expanding markets. Corporately, the company specializes in processing, analyzing, capturing and archiving audio, video and other rich-media content. Government customers include the U.S. Army and the CIA.
As its entry into the security industry, Sonic Foundry is offering an authentication solution called Unified Security View (USV), which is based on seven years of research at Carnegie Mellon University — a $20 million project funded by government agencies and private corporations such as NASA, DARPA, Boeing, CNN, Intel and Microsoft. The system is an integrated multi-pronged means of capturing, analyzing, accessing and managing various types of media and biometric information. It is based on ISLIP (Integrated Speech, Language and Image Processing), a proprietary indexing and analysis engine. The system analyzes various types of biometric measurements and provides an accept or reject decision based on a combination of the results of the various measurements.
“We don't think you need a national ID,” says Krishna Pendyala, senior vice president for strategy for Sonic Foundry. In the proposed authentication scenario, a card number is used as a “key” to unlock the biometric information from the secured database. It is a process of “logically attaching” the biometric information to existing ID cards, says Pendyala.
Enrolling users into the system could be simple and automatic — standing for 15 seconds or less in front of a reception desk could provide multiple video and audio samples that could be used to enroll in the system, or to compare to information from a previous enrollment to verify identity. Off-the-shelf components are used to capture video, voice and even iris scans or fingerprints. The power of the system is in the software used to capture and analyze the data and to make decisions based on multiple biometric features, says Pendyala. Even if no one biometric measurement is conclusive, combining multiple biometrics with the software analysis tools — so-called cognitive technology — can ensure a good match.
Privacy issues relating to the technology are a non-starter, says Pendyala. The biometric information stored in the secure system is nothing but a set of mathematical representations of the real things — they would be of no use to a criminal or even to a sneaky marketer. False rejects and false accepts — standards by which biometrics are generally measured — have not been studied as they relate to multiple biometric measurements, says Pendyala, but the National Institute of Standards and Technology (NIST) will be undertaking such evaluations in the near future.
The technology will be especially useful for applications involving a large numbers of people with an existing system of ID cards, according to the company.
