Communications of the ACM
Two Fraudsters, One Passport – Computers Detect Fraudulent Identity Photos
Researchers at the University of Lincoln in the U.K. found that computers are more accurate than humans at detecting photos that have undergone face morphing, a kind of digital manipulation in which the images of two people are merged together to sufficiently resemble both people. This fraudulent image can then be submitted as part of the application for a genuine passport or driver's license, potentially allowing both people to use the same genuine identification document.
The researchers fed face-morphed photographs into a simple computer algorithm trained to differentiate between morphs and normal photos, and found that the algorithm was able to correctly identify 68% of the morphed images; human participants correctly identified face-morphed photos about 50% of the time. "Our research . . . suggests that the use of computer algorithms may be a better method for minimizing how often these kinds of morphing attacks slip through the net," says University of Lincoln researcher Robin Kramer.
From University of Lincoln
View Full Article
Abstracts Copyright © 2019 SmithBucklin, Washington, DC, USA
No entries found