Iris Recognition in Visible Spectrum by Improving Iris Image Segmentation
Purvik N. Rana, Krupa N. Jariwala · Journal of Emerging Technologies and Innovative Research · 2015
Iris recognition system of the biometrics discriminate the human being by the unique characteristics of their individuality. Iris Image segmentation, as a part of the iris recognition procedure requires higher level of accuracy and effectiveness to achieve greater security for the system. Additionally, eye images captured in visible spectrum need special segmentation care then NIR bases eye image. Iris segmentation highly depends on the acquire image condition. To achieve this, we have calculate improve results for iris segmentation that consists retinex based contrast enhancement for better accuracy. Isophote curvature base eye center localization is implemented to define region of interests with canny edge detection and circular hough transform to accurately locate the boundaries of iris and pupil. Morphological operation base eyelash detection steps are also implemented which regularize final segmentation results. Experimental results on the sample images of UBIRIS database reveals effectiveness in segmentation as well as recognition rate. It fails on few iris images consider to be resolve in future work. From performance evaluation, it seems clearly that proposed iris segmentation has achieved greater level of recognition rate with effective FAR and FRR for sample sub test dataset.