2017/10/14 by Avantika Singh, Vishesh Mistry, Singh, Avantika +5 · 1 citation
Computer Science · Medicine · #Biometric Identification and Security #Glaucoma and retinal disorders #Intraocular Surgery and Lenses
paper · pdf · doi:10.48550/arxiv.1710.05152
Iris serves as one of the best biometric modality owing to its complex,\nunique and stable structure. However, it can still be spoofed using fabricated\neyeballs and contact lens. Accurate identification of contact lens is must for\nreliable performance of any biometric authentication system based on this\nmodality. In this paper, we present a novel approach for detecting contact lens\nusing a Generalized Hierarchically tuned Contact Lens detection Network\n(GHCLNet) . We have proposed hierarchical architecture for three class oculus\nclassification namely: no lens, soft lens and cosmetic lens. Our network\narchitecture is inspired by ResNet-50 model. This network works on raw input\niris images without any pre-processing and segmentation requirement and this is\none of its prodigious strength. We have performed extensive experimentation on\ntwo publicly available data-sets namely: 1)IIIT-D 2)ND and on IIT-K data-set\n(not publicly available) to ensure the generalizability of our network. The\nproposed architecture results are quite promising and outperforms the available\nstate-of-the-art lens detection algorithms.\n