UK academics have developed a unique identification system that encodes facial characteristics into packets of data small enough to fit onto a credit card.
The technique, developed by researchers at the University of Kent, digitises and stores facial components in just 50bytes of data, rather than compressing the image of a whole face. When stored on a magnetic stripe or chip card, the data can be reconstructed to verify a card holders' identity.
Retailers can compare the data on a card with a library of pre-programmed facial characteristics stored on a card-reading device. By tallying card-holders' facial components with the images on the database, users can accurately reconstruct a face and verify a person's authenticity.
Researchers created the face databank by sampling nearly 300 local students and assigning code to each characteristic. The human face can be broken down into 50 to 100 different visual components, said Chris Solomon, who heads up the project.
Mat Lab?s Mat Lab mathematical software, more commonly used for heavy-duty analysis and number-crunching, helped Solomon's team build the complex program. The system could also be used with closed circuit cameras at automatic teller machines and other sensitive access points to improve security.
A dozen European companies have teamed up with the University to help exploit, and gain EC funding for, the new development.
Connexin drops out of Ofcom auction due to start next week
SwiftKey users now send two billion emoji every week
Recruitment plans are 'most ambitious ever', claims Openreach HR director Kevin Brady
Samsung's under-the-hood improvements separate the S9 from the pack when it comes to the display