Researchers crack iris recognition

An international team of researchers has devised a method of reverse engineering code stored in biometric databases to fool iris recognition systems.

Linked gallery: Photos: BlackHat 2012

Iris recognition systems are currently deployed by corporations and law enforcement entities around the world, including at Amsterdam's Schiphol Airport and Google's data centres.

Such systems typically scan individuals' irises to produce code that is then filed in a database and used for future matching.

Professor Javier Galbally of the Universidad Autonoma de Madrid told the Black Hat conference last week that his team had developed a genetic algorithm to reproduce images of individuals' irises by reverse engineering the database code.

Australia has yet to introduce privacy laws to govern the use and storage of biometric information, despite moves to revoke an unsuccessful, voluntary industry code (pdf) earlier this year.

Hackers could then fool security systems by printing the image out to be scanned by the recognition system, for example, by patching the image onto a contact lens to be worn by the attacker.

Galbally said the iris provided among the most reliable forms of identification -- even better than fingerprints -- but "the main problem with the iris is the acquisition".

"Sensors are more expensive, and it's more difficult to acquire because you need more cooperation from the users," he noted.

"The commercial [iris] system only looks for the iris [code] and not an actual eye."

Galbally said there had not been any breaches reported as a result of a bypassing iris recognition systems through synthetic iris images.

"You never know if it's going to be dangerous or not, but the vulnerability is there," he said. "It's good that people are aware that these vulnerabilities exist."

Galbally's research was done in partnership with the West Virginia University in the US and the Biometric Recognition Group-ATVS.

This article originally appeared at scmagazineus.com