Contactless fingerprint biometrics accuracy improves with multiple fingers, NIST report shows
A new report on the accuracy and interoperability of contactless biometric fingerprint scanners has been published by The National Institute of Standards and Technology (NIST), assessing six leading commercial devices. The agency concludes that contactless fingerprint biometrics are currently not as accurate as contact scanners for matching captured prints to images in a database, but the ability of contactless devices to scan multiple fingers on a hand improves their performance.
Accuracy with contactless devices matching a single finger was relatively poor, at only 60 to 70 percent, according to the report, but one contactless system matching multiple fingers performed within a fraction of a percent of the 99.5 percent accuracy that is the baseline for most contact scanners. A mobile app achieved 95 percent accuracy, and other devices were close to 90 percent accuracy. All contactless systems provided low false positive match rates, further supporting their potential.
The agency collected fingerprints from 200 volunteers in a laboratory setting using all six touchless devices and two state-of-the-art contact devices for comparison. NIST previously published a report on contactless fingerprints in 2018. Four of the touchless systems were based on mobile apps, and two on stand-alone devices.
Results in the 48-page report, titled “Interoperability Assessment 2019: Contactless-to-Contact Fingerprint Capture” are anonymized, though all were commercially available as of May 2019. Companies offering contactless fingerprint technology include Idemia and Veridium, and the market for contactless biometrics is forecast to skyrocket to $70 billion by 2030.
The technology is expected to be in greater demand due to hygiene concerns around shared surfaces like contact fingerprint scanners, though NIST says the study began before the novel coronavirus appeared.
Contactless fingerprinting is significantly different in nature from touch fingerprinting, the agency notes, explaining that contactless systems work by illuminating the 3D surface of the finger to create a model through reflection and shadow. Standard fingerprint databases are based on the 2D images captured by touch systems, which tightly couple the images with print features. Evaluating the ability of touchless systems to work with legacy databases was part of the motivation for the study.
“The report summarizes the state of the art of contactless fingerprint scanning,” said John Libert, one of the report’s authors, in the report announcement. “It can help anyone interested in adopting contactless technology to evaluate the cost in performance they might pay by switching to contactless fingerprint capture.”
NIST is planning to release a more detailed follow-up analysis of the data gathered, and a detailed procedure for testing contactless devices later this year.
“One purpose of the research was to further test the hypothesis that multiple finger matching can substantially improve the accuracy of contactless fingerprint matching,” Libert adds. “Our data suggest that it can.”
NIST Biometric Standards and Testing Lead Patrick Grother told Biometric Update in a recent interview that the agency will also look into the effectiveness of facial recognition systems with masks in the near future.