Texture removal for adaptive level set based iris segmentation
Xiaobo Zhang, Zhenan Sun, Tieniu Tan · 2010
Level set based active contour method has been proposed for iris segmentation in recent years, but it can not converge to iris contours in real applications because of its sensitivity to local gradient extremes due to the complex iris texture. In this paper, a novel scheme is proposed to remove local gradient extremes before using level set directly. Firstly, we use two orthogonal ordinal filters to obtain robust gradient map. Then we localize the iris region on the gradient map by an improved Hough transform. After that, a Semantic Iris Contour Map is generated by combining the spatial information of coarse iris location and the gradient map as the edge indicator for level set segmentation. For robust and accurate segmentation, we propose a convergence criterion and a means of updating the parameters for level set. Finally, the accurate segmentation is obtained by the robust adaptive level set method. Encouraging results on ICE 2005 database and CASIA v3 database show the efficiency and effectiveness of our method.