Private Eyes Are Watching You

May 19, 2008 — -- A 17-year-old walks into a liquor store, carries a 12-pack of beer up to the counter and hands the clerk a flawless fake ID. Unbeknown to him, the clerk need not even glance at the ID before turning him down. His face gave him away.

A facial recognition system placed behind the store counter analyzes the teen's 17-year-old features and informs the clerk of his illegal age. It's just one of a litany of uses for the fast-evolving surveillance technology, a field that has security experts salivating and privacy advocates bracing for a battle.

Computers that can pick out fugitives in a crowd, video cameras that scold people for littering, eyes in the sky that detect crimes as they're being committed. While these scenarios may sound straight out of George Orwell's "1984," they are becoming reality and could be headed for your corner store sooner than you think.

Although still being researched across the globe, facial recognition technology has already taking hold, particularly in Great Britain.

Last week, Budgens, a U.K. grocery story chain, announced that it would use facial recognition technology to prevent its clerks from selling alcohol and cigarettes to underage customers. The photos of customers who were refused previously will be stored in a database, and then if the offenders come in to buy similar products again, the clerk will be alerted.

Similarly, the British government plans to roll out a facial recognition pilot program in London airports this summer. People who hold biometric U.K. and EU passports can pass through unmanned gates. At the gate, their faces will be scanned to match them to their passport records.

Though the technology has been around for years and the British are embracing it and moving forward, technology experts say facial recognition -- and the cameras needed to support it -- wouldn't fly with privacy-obsessed Americans, at least not yet.

"One of the reasons that the United Kingdom. has moved forward on this a little faster than other places is there's just a large use of cameras in support of crime reduction in general," said Aaron Bobick, chairman of the school of interactive computing at Georgia Tech, which has done research into both facial and gait recognition. "Once that starts to make video available, then the question of what you can do with that video becomes important."

Big Brother Britain

Britain has a long history of watching its citizenry. For the past 25 years, the public places in many U.K. cities have been monitored by closed circuit television cameras, or CCTV. The surveillance cameras were prompted partially by IRA bombings and terrorism fears.

According to a 2006 report by the Surveillance Studies Network, a nonprofit watchdog group, and presented to the Information Commissioner's Office, the United Kingdom has more surveillance cameras than any other in the world; at least 4.2 million of them are pointed on the country's public streets, one camera for every 14 people.

"[Facial recognition] really has picked up steam in the last 10 years," said Vijayakumar Bhagavatula, who teaches electrical and computer engineering at Carnegie Mellon. "The principle has been around for 25 years, but it started getting put into commercial systems five to 10 years ago."

Bhagavatula describes the technology simply.

"Let's say a digital camera is taking a picture of someone's face. So now it gets represented in computers as a bunch of numbers," he said. "Humans have no problem [saying] that's someone I know. The computer has to look at those numbers and say, 'Are these the same set of numbers corresponding to a person I took a photo of a year ago?'"

It's a complex process, and it is not flawless. For computers, those numbers representing human features can change based on the person's expression, lighting and overall quality of the image, according to Bhagavatula.

To combat this, researchers are constantly looking for new algorithms to analyze facial features. Currently, many researchers are looking at features that don't change, such as the distance between the eyes, the angle made by the tip of the nose or the length of an eyebrow, he said.

"Many methods try to capture these kinds of things that are unique to people's faces," he said. "You hope that these numbers stay the same when a person smiles or frowns."

The U.S. Privacy Police

The kind of monitoring that would enable facial recognition to work well has not caught on in the United States, at least not yet, according to Paul Saffo, a technology forecaster in Silicon Valley.

"The English have always had a slightly different attitude toward privacy," Saffo said. "They've never had a strong a privacy culture as America has had. The English do not have a constitution. Their protections are in common law. It is easier for the government to overstep notions of privacy than it would be here, because you have people invoking the Bill of Rights."

But Saffo believes that given the right crisis, the United States would eventually accept the technology.

"Do not underestimate the psychic shock of the London subway bombings," he said. "We bleat and cry about privacy, but we happily surrender our privacy for the cheapest of coin."

So far, most legislative pushes for video monitoring by city governments have been thwarted.

This week in Washington, D.C., a bill pushed by the city's mayor calling for nearly $1 million in funding for citywide public cameras was voted down by the city council.

"People sometimes talk about video surveillance systems as moving forward inexorably in the United States, but we've seen quite a few successful protests," said Mark Rotenberg, the director of the Electronic Privacy Information Center. "I think there are a lot of questions that need to be asked about video surveillance. The most obvious one is: what is the purpose?"

"[Britains] have embraced a really extraordinary amount of monitoring by the government that I don't think the U.S. would accept," he said.

Yeah, but Does It Work?

Some critics also take issue with the accuracy (or lack thereof) of facial recognition technology.

In perfect conditions, facial recognition can be fairly effective, according to experts, but in less than perfect conditions it can be wildly inaccurate. For example, it is difficult for a computer to identify a person who is walking on a city street or in an airport where his face might be blurred, obscured or shadowed.

"We have gotten a long way from where we were 10 years ago," says Carnegie Mellon's Bhagavatula. "But good algorithms have an 80 percent accept rate. It's pretty good, but not perfect."

Rob Jenkins, a psychology professor at the University of Glasgow in Scotland, may have found at least one way around the technology's inaccuracies. Jenkins and his colleague Mike Burton published a study in the journal Science in January that outlined a method to get 100 percent accuracy from computers by using what the researchers called an "averaged" face image, made up of 20 photos.

"The great thing about this averaging process is it just washes out all these differences of single photographs. The lighting and the pose all kind of becomes neutralized," Jenkins told ABCNEWS.com in January. "And what you're just left with is the core of the face. The aspects of the image are consistent from one photo to the next."

Since that study, police, governments and companies have shown interest in his research, Jenkins said. And although he is interested more in how the mind recognizes faces than how the technology is used, as a citizen, he finds the ubiquity of CCTV troubling.

"New technologies that are being unveiled as being the solution to problems -- often they're just a better key to locking and unlocking something, but that doesn't mean that you shouldn't think about what's behind the door," he said. "Because if you put all this trust in a new technology, ... you can find yourself in quite a hairy situation."

Jenkins points out that sometimes even humans can't recognize familiar faces.

"The human brain is the most sophisticated computer we know of," he said. "Engineers are setting themselves [up] with a very difficult problem by demanding accurate performance. Even humans can't do this reliably and should give us pause. … Is the goal a realistic goal? Are we ever going to build a machine that can do that? And maybe we will, but I think it's a question that's worth asking."