Real-World Non-NIR Illumination and Wavelength-Specific Acquisition Variants in Iris Recognition

Abstract

Experiments with real-world non-NIR illumination as well as wavelength-specific acquisition settings are conducted with respect to impact on iris recognition performance, in particular considering variations in genuine score distribution. Illumination includes daylight, light bulbs, fluorescent tubes, halogen lamps, and mixed modes, while acquisition is done in visible wavelength, near infrared wavelength, and a combination of both. Results indicate very different results for blue and brown eyes respectively, with halogen illumination and visible and mixed-mode acquisition being an interesting option for a compromise setting.