The New Face of SurveillanceBy Sean Gallagher | Posted 2004-11-01 Email Print
WEBINAR: On-demand webcast
Next-Generation Applications Require the Power and Performance of Next-Generation Workstations REGISTER >
Matching mugs may help cops catch crooks, but it could send your customers on the lam.
You're a passenger in a car that gets pulled over for a traffic violation. The policeman looks at you suspiciously and asks your name. He snaps a picture of you with a digital camera and walks back to his cruiser.
A few minutes later, you're spread-eagled on the hood of the car as he cuffs you. Why? He's matched you up with a wanted felon in a databank of faces back at the county jail.
While the odds are you won't be caught in this nightmare anytime soon, the proliferation of facial-identification systems is making encounters like this more likely by the day.
In Pinellas County, Fla., sheriff's deputies armed with Hewlett-Packard digital cameras and laptops connected to a wireless data network can use software from Viisage to check the identity of suspects in a minute or less. The software was credited in a September arrest of a woman who was wanted on two felony warrants.
Mohamed Lazzouni, the chief technology officer of Viisage, says that the query template, a mathematical representation of the captured facial image, is only about 1 or 2 kilobytes. That means the response from the databank, which runs on an Oracle database on a Windows server back at the jail, is essentially real-time. Results are sent to the patrol car in about a minute, and include up to 20 possible matches.
Facial-recognition systems are, of course, only as good as the pictures they're based on and the people who use them. Software like Viisage's package tries to prevent "false positive" matches by bringing back a set of the most likely matches and leaving it up to the user to make the connection between a person and a record in a facial databank.
"Yes, you can ask a question and get a false positive," says Lazzouni. "But the software never returns just a single hit." This means, in theory, that the blame for a bad match falls on the shoulders of the cop, a.k.a. the user.
On the flip side of the coin, it's all too likely that your facial data will be accurately matched some day. As facial recognition becomes a bigger part of what state, local and federal agencies (and companies) do to keep identities straight, personal privacy is at risk. You could find your face giving away everything from your tax payment status to your purchasing patterns.
While biometric systems like fingerprint scanners only record a mathematical model of your identifying characteristics, facial-recognition systems depend on an image that can be viewed by a human being, not just a computer, to confirm identity. That means it's much less difficult to cross-reference digital storehouses of faces, because you can reprocess the raw images with whatever mathematics each database uses to profile the contours captured in digits. And soon there may not be a lot of cross-referencing, as products emerge as de facto standards.
Pinellas County shares its Viisage databank of more than 700,000 previous offenders with other Florida jurisdictions that have bought Viisage software. They return the favor. Other states are tracking the faces of every licensed driver to prevent identity fraud. And as Baseline reported earlier this year ("What Sin City Can Teach Tom Ridge," April, p. 32), Las Vegas casinos share their facial-image data of "undesirables" with each other to keep them away from the gaming tables.
The cost of the technology for recognizing individuals' faces will drop. Viisage and other companies are working to put the technology onto a single chip. So those cameras embedded in card tables that have made poker a TV sport could be watching a lot more than what a player holds in his hand.
In fact, being able to identify customers from video feeds could become a powerful customer relationship management tool. Banks, clothing retailers and even grocers would want to know which customers have walked into their stores.
But before you start thinking about how that sort of information could help your bottom line, think about what it could do to your relationship with those customers.
Does Mr. Jones really want you to know he lingered at the adult-magazine rack before he took his bag of kitty litter to checkout? Or will that kind of information chill your relationship with your customer, rather than make it closer?