Neurotech Scores High in NIST’s Proprietary Fingerprint Model Assessment


Neurotech says it has retained its top score in the National Institute of Standards and Technology (NIST) Proprietary Fingerprint Template (PFT) III biometric performance assessment results, continuing its algorithmic prowess for fingerprint templates since its submission of 2019.

the assessment results which were published in March 2022 examine the capabilities of Neurotechnology’s biometric algorithm for one-to-one fingerprint verification and the accuracy of end-stage fingerprint matchers used in the system’s one-to-many searches Automated Fingerprint Identification System (AFIS) versus proprietary models. It folds into the PFT II test, which is an individual fingerprint verification assessment that examines categories such as plain vs. plain, plain vs. rolled, and pattern sizes; and PFT III, a continuation of previous PFT assessments but with new datasets. Neurotechnology says the March assessment also tested the 2003 PFT standards.

To evaluate Neurotechnology’s MegaMatcher algorithms, 22 vendors submitted 39 fingerprint recognition systems to NIST that were tested 33 times against fingerprint databases held by law enforcement and border agencies. The combined data sets from the Arizona Department of Public Safety (AZDPS) and the Los Angeles County Sheriff’s Department (LASD) tested from one to nine for single flat, from 10 to 18 for single to roll and 19 to 27 for roll to roll fingerprint match. The Department of Homeland Security (DHS2) dataset rated 28 to 30 experiments for clear-text fingerprint matching. The combined biometric datasets of US VISIT POE data with Bio-Visa applications (POE + BVA) were used to perform clear fingerprint matching tests.

Its performance in terms of false mismatch rate to false match rate was the best among bidders, with the lowest detection error trade-off curves for all fingers, according to the announcement.

However, the average model creation time was 228.2 milliseconds without failure, compared to 204 milliseconds in 2019 and the average model comparison time was 4.6 milliseconds compared to 3.7 milliseconds in 2019, which shows a speed slightly reduced.

Evaldas Borcovas, Head of Biometrics Research at Neurotechnology, says of the assessments, “It’s hard to continually improve your technology when it already outperforms all others in the industry, but our team has continued to demonstrate “a spirit of innovation, and we’re excited to see the latest algorithm improve our high scores even further. NIST is considered the most influential and respected organization evaluating biometric algorithms, and their highest rating of our technologies shows that our fingerprint algorithm is the most reliable and accurate algorithm available.

In 2019, the Lithuanian company won the crown of the MINEX III (Minutiae Interoperability Exchange) assessments for biometric fingerprint algorithms. Most recently, in February 2022, Neurotech recorded significant improvements for its facial biometrics in NIST’s two major FRVT assessments.

Article topics

precision | algorithms | biometric identification | biometric matching | biometric tests | biometrics | fingerprint recognition | Neurotechnology | NIST

Source link


Comments are closed.