Iris boundary localization based on Hough transform and the quadratic circle data compensation

摘要

Iris localization is the crucial link of iris recognition and automatic eye tracking. Based on the traditional Hough transform, this paper proposes an accurate pupil detection method combined with ellipse fitting and circular data compensation. We used the minimum gray mean method to approximately determine the inner edge. According to the results, the inner edge image is extracted and finely located by the center compensation method. When locating the outer boundary, a coarse localization is performed based on approximate radius compensation. The fine localization of the outer edge is performed based on approximate circle center compensation. The experimental results show the proposed algorithm improves the accuracy and real-time performance of the localization compared with the traditional method. It retains the original advantages of Hough transform while reduces the amount of computation and useless information.