Abstract:

The existing iris recognition algorithms are of low executing speed and poor robustness. In order to im- prove the accuracy of iris recognition, a novel preprocessing approach is proposed according to the characteristics of iris images. In this approach, first, the least square method is used to locate the inner iris boundary and to test the eyelid. Next, the improved randomized Hough transform is adopted to locate the outer iris boundary. Then, the double-threshold method is employed to detect the eyelash. Finally, a normalizing, de-noising and enhancing process is performed for the circular iris image. The iris image after the preprocessing basically eliminates the dis- turbance of eyelid and eyelash, thus facilitating the following extraction and matching of iris features. Experimental results demonstrate that the proposed approach effectively overcomes the long runtime and the poor generality in image preprocessing, and improves the stability and accuracy of iris recognition system.

Key words: image preprocessing, iris recognition, iris localization, disturbance detection