Iris spoofing detection using discrete orthogonal moments

Abstract

Human iris being the most stable biometric modality suffers from presentation attacks like colored textured contact lenses and print attacks that obfuscate the natural iris texture. The paper presents discrete orthogonal moment-based invariant feature-set comprising of Tchebichef, Krawtchouk and Dual-Hahn moments which are extracted at localized iris regions to capture local intensity distributions of the iris texture. The orthogonal moment-based feature-set is made rotation, translation and scale-invariant in order to accommodate for geometric transformations when images are acquired in uncontrolled environment. The performance of the proposed techniques is evaluated using four publicly available iris spoofing databases: IIITD-Contact Lens Iris, IIITD Iris Spoofing, Clarkson LivDet 2015 and Warsaw LivDet 2015. The textured contact lens detection rate of 100% for IIITD-CLI and 99.48% for Clarkson datasets is achieved, respectively. Similarly, print+scan and print+capture attacks are detected with 99% and 98.93% accuracy for IIS datasets, respectively. The print attacks are detected with 99.63% and 98.89% accuracy for Clarkson and Warsaw datasets, respectively. The proposed techniques thus, prove to be effective in terms of contact lens and print attacks detection when acquired using multiple sensors.