Nonlinear Iris Deformation Correction Based on Gaussian Model

Current iris recognition systems can achieve high level of success under restricted conditions, while they still face challenges of utilizing images with heavy deformation caused by illumination variations. Developing methods to alleviate the deformation becomes a necessity, since the requirement of uniform lighting is often not practical. This paper introduces a novel algorithm to counteract elastic iris deformation. In the proposed algorithm, for nonlinear iris stretch, the distance of any point in the iris region to the pupil boundary is assumed to be the corresponding distance under linear stretch plus an additive deviation. Gaussian function is employed to model the deviation. Experimental results on two databases with nonlinear deformation demonstrate the effectiveness of the algorithm. The proposed iris deformation correction algorithm achieves a lower Equal Error Rate (EER), compared to the other two linear and nonlinear normalization methods in the literature, making the system more robust in realistic environments.