Sources: A. Czajka, P. Maciejewicz and M. TrokielewiczFollowing news from last year that a group of hackers, called the Chaos Computer Club (CCC), successfully hacked a smartphone’s iris-recognition security feature using an artificial eye, a team from the Warsaw University of Technology in Poland has been working to find a solution to the security threat posed by such a breach.

“The security risk to the user from iris recognition is even bigger than with fingerprints as we expose our irises a lot. Under some circumstances, a high-resolution picture from the internet is sufficient to capture an iris,“ said CCC spokesman Dick Engling in a blog regarding the hack.

Working to resolve the issue, the team, led by Mateusz Trokielewicz, trained a machine-learning algorithm to scan a database with iris images from both living and deceased subjects. Once trained, the algorithm could distinguish between the irises of those living and deceased with 99 percent accuracy.

To ensure accuracy, the algorithm was trained on almost 575 near-infrared iris images from 17 different individuals at various stages in the days and weeks after death (from five days after death to 34 days after death). Those images were then compared with 256 images of live irises.

“No post-mortem sample gets mistakenly classified as a live one, with a probability of misclassifying a live sample as a dead one being around 1 percent,” according to the paper.

Unfortunately, for now, the research still doesn’t specifically address what prompted the study in the first place: artificial irises. Yet, the team believes that if the system can distinguish between a living iris and a deceased one, then the algorithm would more than likely be capable of telling a fake iris from a real one.

The research is published on Arxiv.org.