High confidence personal identification by rapid video analysis of iris texture

Abstract

The concept of using the iris of the eye as a kind of optical fingerprint for personal identification is discussed. By isolating and encoding the visible texture of the iris from a video image into a multiscale sequence of quadrature 2D Gabor coefficients, a 256-byte iris code has been developed that uniquely identifies any individual, with astronomic confidence levels. Following image analysis, the critical problem of pattern recognition is transformed essentially into a statistical test of independence on the real and imaginary parts of the complex 2D Gabor coefficients. Statistical decision theory permits rigorous execution of identification decisions from comparisons of iris codes, at the rate of 4000 per second, including computation of the confidence associated with each identification decision. Such a recognition system has been developed using a conventional zoom videocamera. In tests to date, the system has never failed to identify any enrolled individual correctly.<>