To test the accuracy of a new fingerprint scanner, researchers typically run millions of known fingerprint images through the system's matching software. But this testing procedure can't quite mimic real operating conditions, as a 2-D image fed into a program is fundamentally different than a 3-D finger pressed to a sensor.
To get around that problem, researchers at Michigan State University and the National Institute of Standards and Technology have come up with the first 3-D-printed fingerprint. A new technical paper describes their system for projecting 2-D images onto a generic 3-D finger surface, then fabricating the realistic 3-D fingerprint, with all its loops and swirls, in a commercial 3-D printer.
This could be useful for end-to-end evaluations of fingerprint matching systems, which start with fingerprint image acquisition and then go on to feature extraction and matching. In the video below, MSU professor Anil Jain says the use of such 3-D fingerprints could help both sensor manufacturers and algorithm developers improve the hardware and software of fingerprint matching systems. The dummy fingers will also be helpful in the development of up-and-coming touchless fingerprint sensing technologies.
Fingerprint biometrics are finding more and more applications in our speedy and security-conscious world; the uses go far beyond law enforcement. In India, the government is trying to enroll every citizen in a biometric ID system using fingerprints and iris scans. Fingerprints are recorded at many nations' border crossings. And the iPhone 5s can be unlocked with Touch ID, a fingerprint recognition system. So long as our fingerprints are going to be scanned everywhere, we can at least root for those scanners to be as accurate as possible.
Eliza Strickland is a senior editor at IEEE Spectrum, where she covers AI, biomedical engineering, and other topics. She holds a master’s degree in journalism from Columbia University.
