A Novel Iris Location Method for Fast Iris Recognition
Linhua Jiang, Ying Zhang, Wei Li · 2009
Iris recognition is of great significance in biometric technologies, due to the highest reliability and accuracy. As the first crucial step of iris recognition, iris segmentation is continuously and widely studied. A novel iris location method, based on rotation average analysis of intensity-inversed image, is presented here for iris segmentation: First, it locates the pupil by checking the consecutive high intensities in the inversed image, and then refines the center with the centroid of the pupil. A rotation average curve of the centered inversed image is thus calculated and segmented to locate the boundaries of iris. Finally, a fast and precise refinement of the outer boundary of iris is carried out. Both self-collected data and certain iris image database were used in the experiments. Compared with conventional methods, this novel method shows faster and robust properties in locating the pupillary and limbus boundaries with higher tolerance of faintness and scaling.