Modeling of Iris Structure for Homeland Security Applications (MISHA)

We propose to develop a physics-based framework for modeling iris texture, image acquisition conditions (in the broad sense) and other real-world effects related to iris biometrics. The framework is coupled to a special statistical procedure that allows generation of large-scale samples of iris images representative of a desired ethnic group. Our approach is based on the physics-based iris modeling approach available in the literature, extended and modified as appropriate. The main proposed innovation is special estimation and goodness-of-fit testing procedures that allow tailoring the parameters of the physics-based iris simulation scheme to specific iris image/s from a given database. The distinctive advantages of our modeling tool to be developed over Phases I and II, are the unprecedented level of realism of synthetic images, the ability to generate a large-scale sample representative of a target demographic population, and the ability to separately include numerous real-world effects. The market for such a product includes government agencies (DHS, FBI, NIST), academic organizations (biometric research) and commercial entities (developers of iris recognition algorithms and developers of iris recognition hardware systems).