IREX III - Performance of Iris Identification Algorithms
Patrick J. Grother, George W. Quinn, James R. Matey, Mei L. Ngan, Wayne J. Salamon, Gregory P. Fiumara, Craig I. Watson
Iris recognition has long been held as an accurate and fast biometric. In the first public evaluation of one-to-many iris identification technologies, this third activity in the Iris Exchange (IREX) program has measured the core algorithmic efficacy and duration of the core processing functions of 92 algorithms from 11 implementing organizations operating on nearly 6 million images of 4 million eyes of 2 million people. As such this report documents the state-of-the art of iris recognition technology operating on archival imagery, demonstrates the existence of a large and diverse range of implementations beyond those described in the academic literature, and identifies factors contributing to recognition failure.