Research on Iris Localization Algorithms

摘要

Iris localization affects the iris texture feature extraction, which has an important influence on the result of the analysis. The inner and outer edges of iris as concentric circles in the traditional algorithm caused low recognition accuracy. The pupil iris and iris-sclera borders are non-concentric circles, so the new localization algorithms were applied respectively. The iris image was carried on the dynamic threshold segmentation, then, point Hough transform to the images was applied to get the center of the circle and radius of the inner boundary. The improved Daugman algorithms was used to search the center of the circle and radius of the outer boundary to get the outer edge within the valid range. Through the test of iris images regardless of different size, different degree of clarity and illumination intensity, the results show that rather good segmentation results can be gotten by applying the improved algorithm.