Biometric-at-a-distance is not here yet

Published 11 March 2008

For four years, the government has poured a lot of money on long-distance surveillance systems which would identify individuals from a distance in a crowd; the technology is not here yet

The New York Times’s Jim Dwyer visited a biometric laboratory and showroom in Lower Manhattan operated by International Biometrics Group (IBG). The lab tests different biometric technologies. Dwyer writes that few biometrics have been chased as hard and — so far — as futilely as systems that can recognize faces from long distances, with the Department of Defense investing hundreds of millions of dollars in them in recent years. If they worked, such systems could look over crowds of people, or just one person at a time, and see if a face matched any image in a database. This is described as “mass covert data capture” in the world of biometrics, but the real world is not yet cooperating. The small bomb set off in Times Square early last Thursday shows why. Surveillance cameras at Times Square apparently did track part of the route taken by a hooded bicyclist suspected in the attack, but even if one of the cameras caught the bicyclist’s face, “you’re probably not going to get a useful match,” said Raj Nanvati, a partner in International Biometrics, which evaluates technologies and advises law enforcement agencies under a contract from the National Institute of Justice.

Casinos can get good pictures of their customers, typically when they check in, and then can track down banned card-counters. Outside controlled circumstances, the routine chaos of daily life makes those kinds of matches a long shot. A recent study in a train station in Mainz, Germany, found that even under the best conditions — the subjects were looking toward high-quality cameras — the systems could match them to the right face in a database only about 60 percent of the time. When the station was dimmer, the matches dropped way off. “For four years, the government has poured a lot of money on long-distance surveillance systems,” IBG’s Theo Cushing said. “They’re starting to focus on things that actually work. But they’re not giving up on the novel concepts.”

One well-known biometric technology — the iris scan — has proved itself to be reliable, Cushing said, but it can be done only under limited conditions. So it is now being used at some major airports for regular travelers. “The holy grail is to do iris scanning at a distance,” Cushing said. One company expects to bring out an iris scanner this year that can read an eyeball from ten feet or so. The manufacturers of iris readers report error rates of under 1 percent, but Cushing said that such claims would not hold up in a wide range of circumstances — including when people with common eye problems like cataracts are included. The reliability improves when two systems are combined, said Peter Cheesman of International Biometrics.

Another biometric technique, vein recognition, peers at the blood vessels in the hand — they are not very deep — and sees deoxygenated hemoglobin in patterns specific enough that some banks in Japan now allow their customers to have their hands scanned instead of requiring PIN numbers. Inexpensive fingerprint readers can be used on home computers instead of a password. Banks are interested in biometrics that could tip them off about the moods of people waiting in line, such as an increase in body temperature, Cushing said. The spread of surveillance machinery has been answered in some places by people not happy about what they see as intrusive monitoring. “Fashion designers are starting design wraps for the head that obscure the face,” Cushing said. “In England, more people are wearing hoodies.”