The US Customs and Border Protection (CBP) at JFK International Airport has started using facial recognition technology to match travellers' faces to the photo on their passport.Vision-Box is providing the core biometric technology, the vb e-pass desktop, as part of a facial recognition solution provided to CBP by Unisys Corporation, for this initial deployment at three JFK arrival terminals. This deployment follows the successful testing of the same system at Washington's Dulles International Airport.According to Vision-Box, the solution supports a simple process: After authenticating and reading the chip of the traveller's ePassport, the vb e-pass desktop takes a biometric quality photo of the passenger's face, matches the captured facial image against the photo stored in the ePassport, and gives a probability match based on multiple facial features. If the image match probability is low, the passenger may be flagged for additional screening."By the end of 2016 every non-expired US passport in circulation will be electronic, while the majority of them already are," says Michael Petrov, Vision-Box North America managing director. "The deployed solution completes the lifecycle loop of the ePassport by providing a reliable means of checking that the passport belongs to the holder, thus helping address a significant security threat of document swapping with unfortunate precedents documented in other parts of the world. It was reported, for example, that a known Australian jihadist was able to leave Australia for the battlefield in Syria using his brother's passport. His travel could have been prevented with the use of vb e-pass desktop."The company says: "The face matching algorithm integrated into the system provides accurate results even when the photograph in the passport's chip is of substandard quality. This algorithm proved itself as a winner of several recent NIST vendor evaluations, and it was further tested during the field trial in the Dulles Airport pilot. Essentially, the system answers the question, 'What's the statistical probability that the document does not belong to the holder?', and ultimately, 'Is this person who they say they are?'