An accurate and easy method towards iris localization

Iris segmentation is an essential module in iris recognition because it defines the effective image region used for subsequent processing such as feature extraction and matching of the iris. Traditional iris segmentation methods often involve in searching large region and the arithmetic is complex. Some suppose that pupil and iris are concentric circles, but most of time, the two circles are not concentric. To address these problems, this paper proposes a new method in iris segmentation. Differences of two histograms are calculated to determine the search distance from the center of the pupil to the upper eyelid. Sable operators and morphological image opening operation are adopted to detect the lower eyelid. Through these operations, the candidate region is relatively perfect and it increases the accuracy of iris location. The gray level changes very quickly at the boundary of iris and this is used to detect these points of the possible iris boundary. To remove the effect of the noise, modified neighbor function criterion algorithm originated from the pattern recognition is adopted. LS (Least Squares) is used for curve fitting. The results on the challenging iris databases (CASIA-IrisV3-Lamp and CASIA-IrisV3-Twins) demonstrate that the algorithm is excellent in both accuracy and speed.