Michael D. Wheeler
A computerized facial recognition system, employing advanced algorithms coupled with off-the-shelf imaging equipment, has given this state's motor vehicles division an advantage in determining a license applicant's identity.
Not long ago, there were few safeguards that allowed the division to verify a driver's identity. A passing resemblance to somebody pictured on an old license and a forged birth certificate were the ticket to a new identity.
To eliminate such abuses, West Virginia has installed a computerized facial recognition system in all of its license issuing offices. In place since October, the systems employ proprietary algorithms that compare unique facial features on a stored digital image with the facial features of a new digitized image. They include the generic shapes of the human face: eyes, nose and general face topography.
The facial recognition system is the result of a collaboration between Polaroid Corp., based in Cambridge, Mass., and VisionSphere Technologies Inc. of Montreal. Polaroid's PC-based image capture station photographs the applicant with a color video camera employing a 2/3-in. charge-coupled device image sensor. After VisionSphere's UnMask software compares the two digital photos, the station prepares the finished driver's license. Each license is laminated to form a bond with the photograph. Any tampering immediately renders it invalid.
Polaroid also has installed other features: fingerprint matching for another degree of screening, along with an optically variable device -- similar to a hologram -- contained in the laminate.
Already in place in 54 locations, the facial recognition system could present problems for those with criminal intent. "The state has told us about brothers who have a basic resemblance trying to get licenses or some who have simply tried to steal a license and doctor it," said Richard Grimm, manager of the West Virginia program for Polaroid. "The system has been fraught with a lot of abuse."
The state hopes the technology will decrease instances of underage alcohol and tobacco purchases, along with certain types of financial fraud. Since it was installed, Grimm said, the department has had an extremely low error rate.
The facial recognition system has already generated interest from other states. The next step could be harnessing the technology for use in certain security applications and in law enforcement.