December 2001

Recognizing the Enemy

Creating a central database of photos to identify terrorists through face recognition is a bureaucratic nightmare.

By Alexandra Stikeman

Of all the dramatic images to emerge in the hours and days following the September 11 attacks, one of the most haunting was a frame from a surveillance-camera video capturing the face of suspected hijacker Mohamed Atta as he passed through an airport metal detector in Portland, ME. Even more chilling to many security experts is the fact that, had the right technology been in place, an image like that might have helped avert the attacks. According to experts, face recognition technology that's already commercially available could have instantly checked the image against photos of suspected terrorists on file with the FBI and other authorities. If a match had been made, the system could have sounded the alarm before the suspect boarded his flight.

In the wake of the attacks, a number of companies, security professionals and government officials have proposed using biometrics-identification based on a person's unique physical characteristics-to enhance airport security. "We've developed some fantastic technologies, but we just haven't deployed them," says Georgia State University aviation safety researcher Rick Charles. Readily available biometric techniques include digital fingerprinting, iris scanning, voice recognition and face recognition.

Of these technologies, face recognition is perhaps the best suited to surveillance of busy public places like airports. For one thing, it doesn't require those being watched to cooperate by looking into an iris scanner or putting a hand on a fingerprint reader; face recognition devices can work with the video feeds from the cameras that are already ubiquitous in public spaces. It's also much easier for authorities to obtain a suspect's photo-from a passport or driver's license, for example-than it is to obtain other biometric identifiers.

Indeed, many government agencies, from the FBI and the CIA to the Immigration and Naturalization Service and the Drug Enforcement Agency, keep large databases of photos. When the FBI places a suspected terrorist on a watch list, the agency circulates that person's photo to local police, immigration officers or customs agents. "But if you're a cop, you've got to be pretty good at quickly scanning faces in a crowd [for terrorists]," says Richard Norton, executive director of the International Biometric Industry Association. With face recognition technology, security officials could link their surveillance cameras to any number of databases via the Internet and let a computer do all the work, alerting officials when it finds a positive match.

Setting up a face-recognition-based security system would be relatively simple. The two major players in the area of face recognition, Littleton, MA's Viisage Technology and Visionics in Jersey City, NJ, say they have systems that could easily do the job. Visionics' device plugs into a video camera and grabs images of faces while the camera is recording. Software extracts the unique characteristics of each face and creates a template, a compressed digital file that can then be sent over the Internet to several databases at once. Further Visionics software installed alongside each database sifts through a million photos per second and signals when it finds a match.