President Bush often warns opponents of the Iraq war that if we weren’t fighting terrorism overseas, we would have to fight it here.
University researchers are working to make sure we’re equipped to do both.
Ramalingam Chellappa, a professor in the A. James Clark School of Engineering’s electrical engineering department, has been working for the past 10 years to develop cameras that can tell one face from another in real time – a technology that could improve airport security and also help troops to recognize their enemy.
“Catching up with the people who want to harm us is a difficult problem,” Chellappa said, “but we’re making progress.”
Last week the university finalized a contract with ACAGI Inc., the company that would market and produce the technology for government and commercial security uses. The Frederick, Md.-based company hopes to have a video recognition prototype ready this winter.
The collaboration is a part of the university’s Maryland Industrial Partnerships program, aimed at funding joint ventures between faculty and Maryland businesses. MIPS will provide more than $109,000 for the research and ACAGI will fund the tuition of one of the doctorate students working with Chellappa.
ACAGI CEO Peter Spatharis said the company sought out the relationship with the university to qualify for state funding, which requires a partnership with the university system.
“It just turned out that Dr. Chellappa is world-renowned in face recognition,” Spatharis said, “so for us it was just a perfect match.”
Spatharis approaches the problem of recognizing enemy forces from an informed perspective. A former member of the Greek special forces, he said he “knew the kind of applications that needed to happen” when he founded the company in 2004.
“The best tool for insurgents is to be able to blend into a crowd,” he said. “You can’t fingerprint a crowd.”
Video recognition uses algorithms to compare individual frames captured by a video camera to photographs stored in a database.
While fingerprinting and optic recognition still have a better rate of accuracy in identifying individuals, Chellappa and Spatharis pointed out that video recognition is “non-cooperative,” meaning soldiers or security personnel don’t have to ask permission or actually detain anyone to know who they are.
“That’s the advantage,” Chellappa said.
He also explained that video recognition could be used to add another layer of security at airports. On the opening slide of a PowerPoint presentation Chellappa has prepared, there is a chilling security camera frame dated Sept. 11, 2001. Mohammed Atta is pictured walking through airport security just hours before he piloted American Airlines Flight 11 into the World Trade Center.
Face recognition cameras, Chellappa said, could help identify known terrorist suspects before they pass through security if airports are provided with a database containing pictures of known-terrorist suspects to make comparisons with.
Actually introducing facial recognition into warfare has proved more difficult, though, he said. The technology has performed well in controlled environments, and Chellappa’s team has solved the problem of stabilizing images from a camera that is attached to a moving soldier. But more development is needed before it’s combat-ready.
And that’s fine with him.
“We just like the difficulty of it,” Chellappa said. “The moment the problems are solved, we lose interest in it.”
Contact reporter Andrew Vanacore at vanacoredbk@gmail.com.
