June 5, 2017 -
At the request of US Customs and Border Patrol, Oak Ridge National Laboratory has developed a new facial recognition system intended for use at land borders that draws on light-field sensor arrays to mitigate the glass barrier of vehicles, according to a report by The Verge.
Current facial recognition systems used at land borders, such as in Hong Kong, require the driver to roll down the window of their vehicle to evade the negative impact that glass has on the system’s accuracy.
The new facial recognition system contains plenoptic cameras (also known as light-field cameras) that are able to focus past the reflections on the windshield to the faces of the drivers and passengers.
If the system is proven effective, it could lead to increased deployment of facial recognition at land borders.
The system is the result of the federal government’s biometric exit initiative, which mandates that all foreign national visitors are biometrically verified as they exit the country.
A GAO report in February revealed that the available technologies used at land borders “would require all passengers to stop and exit their vehicle to be photographed or scanned,” which would create significant delays at some of the country’s busiest crossing points, such as the Mexican border.
CBP approached Oak Ridge National Laboratory last June to develop a solution.
“The camera they have developed can go into a vehicle through tint and glare,” CBP executive director Colleen Manaher said at a conference in April. “I’m looking at it through the naked eye and I can’t see in, but there on that screen are two people, the passenger and the driver, in facial recognition quality.”
Still in the prototype phase, the plenoptic cameras could address two key issues facing in-car facial recognition systems: the glass barrier adds reflections that can negatively affect many algorithms, as well as reduce the total amount of available light.
Plenoptic systems use an array of sensors to maximize the amount of light information captured, as well as provides sufficient information to perceive depth, which is essential to distinguishing glass reflections from the facial features behind it.
The technology is relatively new to government agencies, however, it has been available to consumers since 2012.
“Because it lets you separate objects in space, you can imagine focusing one image on the tinted windshield and another image where you expect the driver to be,” Colvin Pitts, a senior architect at Lytro, said. “With some clever image processing, you could remove some portion of what you get from the windshield.”
It remains to be seen if these techniques would be successful in the field. However, further development of the technology could enable new facial recognition opportunities, particularly when coupled with automatic license plate recognition or ALPR technology.
In March, the CBP updated the industry on its plans for biometric exit since it issued its initial request for information (RFI), CBP OIT Biometric Exit Acquisition last June, in which it emphasized that it will “continue to engage with industry regarding biometric exit.”