Iris identification based on a local analysis of the iris texture

This paper focuses on a new iris identification method based on a local analysis of the iris texture. In the method, the iris is divided in sub-regions, using locally sliding windows, to extract local signatures. Local distances are then calculated and fused, based on a weighting average. The sliding allows to compensate for local distortions due to segmentation imprecision. The applied weights take into account additional knowledge about the information quantity carried by the different sub-regions and its reliability. Tests have been conducted on the CASIA-IrisV3-Interval database. They show good performances of the new method. Similar or even better results are obtained, compared to published ones, with a set of iris containing twice more subjects.