Airport to Get Facial Recognition Technology
A plan to use facial recognition technology at Oakland International Airport has generated ripples of concern among passengers and privacy advocates, and has sent law enforcement officials scrambling to explain their intentions.
The technology, already in place at dozens of police departments throughout the country, will only be used at the airport on individuals who have been arrested by Oakland police, said Deputy Chief Patrick Haw.
But the makers of facial recognition products and other biometric identification systems, such as iris scanners, predict a day when their use will be as commonplace as airport metal detectors or video surveillance cameras at convenience stores.
Already, Boston’s Logan International Airport plans to use the technology on a trial basis in the coming weeks. San Francisco International Airport officials said they also are considering installing the devices but have no firm plans on how extensive the use will be.
The use of similar technology to scan unwitting fans at this year’s Super Bowl in Tampa, Fla., drew a slew of criticism, in part because it was trained on every face that entered the stadium and also because most were unaware of its use.
Using regular video images, the program scans some 200 points on a person’s face and compares them to a database of known criminals or suspects. Images from the database that match the person’s face most closely appear on a video screen and police can compare the stored mug shots to the person in custody.
“It literally picks up the contours of the face, the hollows of the cheeks, shape of the nose and the curvature of the eye sockets,” said Iain Drummond, president and chief executive of Imagis Technologies Inc., the Vancouver, British Columbia, software company that developed the system to be used at the airport. “That information then gets transformed into a digital string, 800 bytes long. It’s the unique digital signature of your face.”
Drummond likens the system to a digital search engine, which uses strings of text to locate information on the Internet. In this case, the search engine is locating images of people with similar facial structures.
Tampa police use a similar system to scan crowds in the Ybor City entertainment district, and it also has been deployed at various casinos throughout the country. Privacy watchdogs say it’s not the technology but its potentially invasive applications they find troublesome.
“The concern we have with most forms of surveillance like this is when they’re used en masse,” said privacy advocate Lee Tien, an attorney for the online civil liberties group Electronic Frontier Foundation in San Francisco. “At the very least, you should know you are going to be scanned.”
Disagreement About Degree of Accuracy
Tien said there’s a difference between the way the technology was used at the Super Bowl and its use to help identify individuals who already are in custody, which is how Oakland police say they will use it at the airport.
“The distinction of using it only after someone has been detained for objective reasons, instead of relying on face recognition to determine who should be detained in the first place, is huge,” he said.
But there are other potentials for misuse, Tien said. “A lot depends on the accuracy or completeness of the database against which you’re matching. The question is, what’s your reliability rate? And who’s in the database in the first place? Do they just have terrorists in it, or is anyone who’s been suspected of a crime? There is always a concern that ‘the computer says so’ will become a self-fulfilling prophecy.”
Because the technology is relatively new, there is broad disagreement about its degree of accuracy. Some experts suggest that it is significantly less reliable when used on a crowd--where people stand at different heights and angles, and at various distances from the camera--than in a controlled setting such as a police department.
Oakland police officials insist they will not be using the system at random. But when a person is arrested and his or her face is scanned, that image becomes a permanent part of the database, Deputy Chief Haw said.
Facial scanning “certainly has promise” for law enforcement, he said. “But we really as a society need to sit down and consider the constitutional and privacy issues at stake, and those are large issues. We have no intention of using this technology that way at the Oakland airport.”
Airport police were slated to receive facial recognition capabilities before the East Coast terrorist attacks Sept. 11. The program, now expected to be implemented by the end of the year, is part of a major airport expansion already underway that includes an increased police presence. Over the last decade, the number of passengers using the airport has nearly doubled, from 5.5 million in 1990 to 10.6 million last year.
Similar Technology Is Used Elsewhere
Oakland police received the technology through the Alameda County Sheriff’s Department, as will all of the other police departments in the county. According to Imagis Technologies, its contract with the Alameda County sheriff is worth $2.65 million. Because the program’s deployment is not complete, sheriff’s officials declined to discuss the technology.
Several other police departments in California also use facial recognition technology, including Chino and Fullerton.
A similar system has been in place at Canada’s Lester B. Pearson International Airport in Toronto for almost a year but toward a different end. Rather than preventing suspected criminals from boarding flights in Toronto, it is used at arrival gates to prevent undesirables from entering the country.
On Wednesday, Amsterdam’s Schiphol Airport began using iris scanners, which register the individual traits of a person’s inner eye, to help priority passengers move quickly through passport control.
In a few years, “we won’t think twice about it,” said Drummond the Imagis Technologies president. He advocates the use of his software at airline ticket counters, where every potential passenger could be scanned before boarding a plane. “It’s not intrusive. It’s not delaying you. There’s no particular reason why the passenger should object,” he said.
That’s not necessarily how the passengers at Oakland saw it on a recent afternoon.
“It would make me uncomfortable,” said Jose Montenegro of Salinas as he waited for his flight at an airport restaurant. “They have to set very specific criteria so people have the means to protect themselves against abuse. They cannot just be profiling people, pulling people aside.”
Greg Silva of Eugene, Ore., agreed. “I think it’s a way of tracking you as an individual,” he said. “In spite of everything that’s going on right now, I think there are too many changes coming too quickly. We fight for our liberties, and then we’re too quick to give them away.”
Philippia Pegram of Oakland said she accepts the idea now, even though it would have troubled her a couple of years ago. “If you have nothing to hide, it shouldn’t concern you,” she said while waiting for her baggage.
Deputy Chief Haw said the system, which is being used elsewhere in the city, has enabled his department to book suspected criminals considerably more quickly, giving officers more time to fight crime.
“It used to be that when people came in after they had been arrested, we took their fingerprints and faxed them to the Sheriff’s Department, and we would wait literally hours to get identification back,” he said. “Once we have a positive ID, we can do a computer search to see if their names are on other systems in different jurisdictions. But the trick is, we’ve got to know who it is we’re talking about.”
Even so, the system has its limitations.
The way the new technology is going to be deployed at the Oakland airport, for example, it would have been useless in stopping Mohamed Atta, the suspected ringleader of the Sept. 11 terrorist hijackings. Unless he had been arrested, he would not have been scanned. And since Atta is not believed to have had a criminal record in Alameda County, his face would not have produced a match.
*
Times wire services contributed to this report.
