L-1 Identity Solutions’ Daugman-based iris algorithm passes test

Published 27 January 2010

Rigorous tests by the U.S. National Institute of Standards and Technology (NIST) find that the accuracy of L-1’s iris recognition technology as markedly higher than prior results, and that improvements came without sacrificing speed; the evaluation also showed that L-1’s accuracy did not come at the expense of template size

Good news for L-1’s iris recognition technology. L-1 Identity Solutions announced that the accuracy, speed, and template compactness of L-1’s Daugman-based iris algorithm was unsurpassed in a NIST (National Institute of Standards and Technology) IREX I (Iris Exchange) supplemental report published this month. The company notes that the NIST IREX I supplemental report adds to the October 2009 release of initial IREX results on the performance of iris recognition algorithms.

The report notes that the accuracy of L-1’s iris technology was markedly higher than prior results and improvements came without sacrificing speed. In fact, when averaged over all tested databases and measured at the most demanding operating point published by NIST (a false match rate of 1 in 1,000,000), the Daugman-based L-1 algorithm produced the best accuracy of all ten iris vendor participants.

The evaluation also showed that L-1’s accuracy did not come at the expense of template size or algorithm speed. Compared to the competition, L-1 templates were ten times smaller than the next most accurate vendor. In NIST’s modeling of a 2,000 enrollee identification for access control (where the time for template creation and matching were both considered), the L-1 iris technology showed the lowest false rejection rate and the fastest composite match time of all five of the most accurate vendors. Among all algorithms participating in the IREX evaluations, the speed of template creation varied over a factor of 100, the speed of matching varied over a factor of 1,000, and L-1 performed at the highest end of both parameters for a given accuracy. As a result, the L-1 algorithm had the best combination of accuracy and speed of all tested algorithms, an attribute that is critically important to facilitating large-scale identity programs and supporting national database applications.

Professor John Daugman, chief scientist of iris recognition for L-1 Identity Solutions said, “Our performance in the NIST IREX test shows that L-1’s iris technology gives consistently accurate matching over a variety of databases without sacrificing template size or matching speed. This translates into huge operational advantages in real world deployments. I am honored to play a role in continuing to set new levels of excellence for iris recognition within the scientific community and together with L-1 remain committed to ensuring that this important technology brings demonstrable and the industry.”

IREX is an umbrella program for various NIST activities supporting interoperable iris biometrics. The IREX I supplemental report is available at NIST Web site.