Local Chan-Vese segmentation for non-ideal visible wavelength iris images

Iris segmentation becomes a challenging task for non-ideal iris shape captured under visible wavelength environment. In this paper, we proposed a localized active contour model for iris segmentation. A Local Chan-Vese (LCV) region-based active contour model is studied and applied to segment the non-ideal visible wavelength iris images. The proposed localized region-based formulation is more robust and suitable in segmenting the papillary/limbic boundary of non-cooperative users under visible wavelength compared to a standard Chan-Vese (CV) segmentation model which has intrinsic limitation when dealing with inhomogeneity properties. We applied B-spline explicit framework to further improve the computational efficiency of the algorithm overcoming the limitations of the original method. Experimental results on NICE.I have indicated good segmentation accuracy using the proposed algorithm.