Tracking Iris Surface Deformation Using Elastic Graph Matching
Shiau Shing Phang, Wageeh Boles, Michael John Collins · QUT ePrints (Queensland University of Technology) · 2006
We propose a method to track the iris surface deformation in image sequences captured by a special infrared illuminated high-speed camera using elastic graph matching. A circular grid elastic graph (iris graph) to track the iris radial and circular movement of pupillary activity caused by varying lighting conditions is introduced. We compare the phase similarity with the magnitude similarity for tracking iris features and we also relate the determination of the weight of topography preservation in the similarity cost function to the amount of movement of the object being tracked. The algorithm is tested with a series of synthetic iris images and a series of real iris images. We show that the deformation of the iris surface area during the constriction of the pupil operates mainly in the middle and the peripheral parts of the iris and that this deformation is non linear.