The segmentation is important issue in iris recognition system. the aim of iris segmentation is to obtain the ROI (region of interest) iris from the eye image. this paper compares two methods of iris segmentation, namely: daugman and hough method. some image preprocessing techniques are applied to improve the performance of these methods. the next step after iris segmentation is normalization process. the normalization transforms the iris image in cartesian coordinates to the polar coordinates. the images are obtained from the normalization will be used in iris features extraction stage. the performances of those methods are tested using eye images from database CASIA 1,2 dan 3 databases. the expriment results showb that the hough method is better than daugman. the best accuaracy of daugman method achieves 92.57%, 26.75%, 56.03% for the CASIA 1, and CASIA 2, CASIA 3 respectively, and the best accuracy of hough method for the same databases is 96.57% 54.5% 94.83% respectively.