Under-Display Ultrasonic Fingerprint Sensor Sees Beyond the Skin, Maps Your Blood Vessels
3D approach could be considerably more secure than current fingerprint scanners, but takes an hour per capture.
A trio of researchers at North Carolina State University have created a fingerprint scanner with a difference: It can peer under your skin at your blood vessels, making it considerably more difficult to fool.
Fingerprint sensors are a common sight on modern smartphones, thanks to their ease of use: Just poke your finger at the sensor and the phone immediately unlocks and is ready for use. Unfortunately, they're also relatively insecure: You leave your fingerprints on every single surface you touch with your ungloved hand, and in 2014 a presentation at the Chaos Communication Congress (C3) demonstrated how a fingerprint can be replicated from a photograph of a waving politician and used to fool commercial scanners.
The solution, according to Chang Peng and colleagues from NCSU: Look beneath the skin, and map out the underlying blood vessels as well as the surface-level fingerprint.
"While fingerprint technology has been widely used in mobile devices for user identification, the existing fingerprint sensors can only capture 2D images of the fingerprint and are thus vulnerable to spoofing attacks using 2D replicas of fingerprint," the researchers explain in the abstract to their paper. "Ultrasonic fingerprint recognition via imaging the structures beneath the human skin can be a promising approach for preventing spoofing attacks on fingerprint-based identification devices.
"In this study, under-display ultrasonic fingerprint recognition (UDUFR) was investigated for the first time via imaging a finger vessel underneath the fingerprint."
To prove the concept, the team built both an ultrasonic sensor and an artificial finger - the latter constructed from two layers of polydimethylsiloxane sandwiching fake blood vessels which were filled with cow's blood to complete the simulation. In testing, the sensor proved able to image both the surface fingerprint and the underlying blood vessels with a 500x500 dots per inch (DPI) resolution — enough for security work.
"We envision this 3D fingerprint recognition approach can be adopted as a highly secure bio-recognition technique for broad applications," Peng tells IEEE Spectrum in an interview on the research, "including consumer electronics, law enforcement, banking and finance, as well as smart homes." There's only one catch: The current system takes a full hour to capture an image of the finger, hence the use of a stand-in for a real volunteer's digit.
The paper has been published for early access in the IEEE Sensors Journalunder closed-access terms.
