Biometrics at DMVs
I posted a story yesterday to SIW about how California is trying to green light a project to use facial recognition software with its driverâ€™s licenses. The technology basically searches the license photos to try to identify similar photos so that the same person canâ€™t get another license in another name. The software has been implemented in a handful of other states, and I would suspect you might see this come to all states within a few years.
Iâ€™m assuming there are competing vendors, but the name that keeps coming to the top is Viisage, which is part of Robert LaPentaâ€™s L-1 Identity Solutions. The solution has landed in such states as Wisconsin and Oregon (see Oregon's public-facing page on this technology) among others, and basically tries to use a one-to-many match based on the measure of key facial features. The software then returns a batch of similar photos from the DMV files that it believes are very similar to the new photo which was taken. Itâ€™s left up to DMV personnel to then compare each of those and decide whether itâ€™s just a very similar face or actually the same person. If the DMV believes that it is the same person, then it goes through a process to verify whether they are unique individuals or whether a license should be revoked. It also can use a one-to-one match, which allows the system to do a real-time scan during the issuance process to see if the new photo for a license renewal looks like the man or woman who was on the last issuance of that license. Apparently it works. By May of 2006, just eight months after Wisconsin deployed such software, they had already revoked 600 licenses.
When I look at the growing deployments of this technology, I instantly think of its impact on the video surveillance and analytics industry, which has always struggled with facial recognition. Why have we struggled? Unlike state DMVs, our images of persons of interest arenâ€™t taken while the person is standing still against a brightly lit, solid color background. However, I can see that actual usage of a DMV facial recognition system will improve the matching algorithms and spur the entire industry toward even greater applications of this technology. With proper placement of cameras and lighting at chokepoints, this has major merit for our industry. And in the economy of scale, once the big guys run out of DMVs to sell this to, itâ€™s only natural that it will be pushed for corporate and government security applications. Such systems are already in use at some casinos, including the Stratosphere Casino in Las Vegas, which uses a Biometrica system.
On a side note, we also posted an article about Oklahoma considering eliminating fingerprinting from the driverâ€™s license and motor vehicle registration process. So as one state moves towards great biometric matching, thereâ€™s always another state potentially going in the opposite direction.
And on a second side note, facial recognition continues to expand into consumer devices, with Panasonic announcing a line of consumer digital cameras that have face recognition built-in. Again: How can all of this (consumer cameras, DMV, etc.) not boost the respectability and quality of face recognition for video surveillance systems?
And on a third side note, 3VR, which includes facial recognition in its searchable video systems, named former Apple executive Al Shipp to its CEO post this week. Ok, thatâ€™s it; I promise no more info on facial recognition in this newsletter!
In other news:
U.S. passport cards hack, Orsus gets IP from Cinario, Alarm systems reduce crime
A â€œwhite hatâ€ style hacker, Chris Paget, exposed this week how easily he could read the U.S. Passport card (not the blue booklet) and enhanced driverâ€™s licenses which used RFID. The Smart Card Alliance was very quick to point out that the technology in the card isnâ€™t the same as the passport booklet, which uses a contactless smart card technology.