Abstract: To improve the speed of localization, a fast iris localization algorithm combined a coarse localization and a fine localization.Firstly, pupil area is segmented from eye image by threshold, removing eyelash noise etc from binary pupil area by morphological operation.Then extracting pupil boundary point by line scanning to the pupil area, and using the least-square fitting to determine the inner boundary.Finally, using the circle gradients operator to localize the inner and outer iris boundary.Average localization time cost by the proposed algorithm on more than 100 iris images from CASIA (Version 1.0) is 1.38s, circle gradient operator cost 9.8s, Hough transform method cost 14.3 s.Experiment results showed that the proposed algorithm has a high performance on speed and precision, also has strong robustness for different quality iris image.