A 3D Iris Scanner from Multiple 2D Visible Light Images

Abstract

The development of new methods for biometrics using the 3-D surface of the iris could be useful in various applications, such as reliable identity verification of people, when only segments of the iris are available, the study of how the iris code changes with pupil dilation, and studying acute angle glaucoma and its relation to the 3-D iris structure. The goal of this paper was to build a 3-D model of the iris surface from several 2-D iris images, adding depth information to the iris model. We developed a 3-D iris scanner, which reconstructs a 3-D mesh model of the iris surface from several 2-D visible light images. First, a smartphone camera captures visible light iris images from different angles in a controlled illumination environment. Then, a structure-from-motion algorithm reconstructs a point-cloud 3-D model. Finally, the best-fitting 3-D mesh model is obtained using the screened Poisson surface reconstruction technique. Our results include the reconstruction of the 3-D iris models of ten subjects. These models contain an average of 11,000 3-D points. The spatial resolution of our device was measured as 11~\mu \text{m} by scanning a 3-D pattern of known dimensions. The 3-D model of the iris is compared with the results from an optical coherence tomography (OCT) performed on one iris. Our results show that our new 3-D iris scanning method produces a model with potential applications in biometrics and ophthalmology.