The US firm NexID Biometrics announced an upgrade to its fake-finger-detection SDK at the three-day Biometrics 2014 conference, at the Queen Elizabeth II Conference Centre, pictured, in Westminster, central London.
NexID said version 1.2 of its SDK, targeted for initial deployments in November 2014, boosts the accuracy rate range to 96.5 to 99.5 percent in catching bogus identification attempts across a variety of fingerprint-sensor types, such as touch or swipe, and technologies, including optical and capacitance. The firm says this was achieved without a significant impact on computing time or resources, thereby maintaining user convenience, the company claims.
NexID Chief Operating Officer Mark Cornett says: “Our ongoing research has yielded additional fingerprint-imaging features that are highly effective in differentiating images captured from authentic versus fake fingerprints. In essence, we have expanded our portfolio of fingerprint-imaging features and that translates into higher accuracy.”
He said the new SDK’s equal error rates – the inverse of accuracy in correctly identifying a spoof image or not — have been reduced by half from the prior version to 0.5 to 3.5 percent.
“We continue to explore ways to reduce equal error rates toward zero so that customers who deploy fingerprint-identification sensors can have complete confidence that authentications of the people using the sensors are accurate. We’re not aware of any software-based, fingerprint spoofing-mitigation technology today that’s more rigorous than NexID’s latest software improvement.”
Last month, NexID announced an overhaul of its software to make its product applicable to fingerprint sensors in mobile devices and embedded modules, such as ATMs and terminals for entry access, time and attendance, and point of sale. The company said then it had redesigned its software for integration with embedded “match-on-chip” implementations where matching and fake-finger-detection software execute on the same processor that controls the fingerprint sensor’s image-capture, enrollment and authentication operations. The firm argues that fake finger detection is now available across the entire spectrum of fingerprint sensor devices.
