Ensemble of Statistically Independent Filters for Robust Contact Lens Detection in Iris Images
Raghavendra Ramachandra, Kiran Bylappa Raja, Christoph Busch · 2014
Contact lenses are known to degrade the performance of the iris recognition system. Thus, accurate detection of a contact lens is of paramount importance not only in improving the reliability but also the security of an iris recognition system. In this paper we present a novel scheme for detecting contact lenses in iris images. The proposed scheme is based on an ensemble of statistically independent filters whose response on the iris image is processed and further classified using a linear Support Vector Machine (SVM) to detect the contact lens. Extensive experiments are carried out on two publicly available large scale databases, namely: IIIT-Delhi Contact lens iris database (IIITD) and Notre Dame cosmetic contact lens database 2012 (ND) that are comprised of contact lens iris samples captured using four different sensors. The rigorous experiments conducted in this work show the outstanding performance of the proposed scheme, especially in detecting a textured contact lens with a Correct Classification Rate (CCR) of 100%. We also present a comprehensive benchmark of the proposed scheme with six different well established state-of-the-art schemes available for the contact lens detection.