A Novel Visible Iris Recognition using Color Monogenic Wavelets

Abstract

Visible light (VL) iris recognition technology[1] has a growing number of applications in the field of security and citizen identification with the growing number of surveillance cameras and smartphone cameras that acquire iris images in the visible wavelength range. The problem of iris recognition has also led to a growing interest in the development of robust algorithms for visible light iris recognition. Visible light iris recognition is also important for better results on light-colored irises which observe a richer capturing of iris patterns in the visible light range as compared to the Near-Infrared (NIR) range.
VL iris images have been shown to have more noise in the form of shadows and reflections and are not as efficient as NIR iris images in capturing iris patterns for dark-colored irises. In this research work, a novel color monogenic wavelength based VL iris feature extraction is performed on widely used VL databases (PolyU[2] and UTIRIS[3]). The results of the proposed algorithm outperform the commonly used algorithms for NIR iris recognition