Iris recognition using SIFT descriptors with different distance measures

Experiments on iris recognition using Scale Invariant Feature Transform (SIFT) features and several distance measures we defined are presented. Two well-known iris databases, UBIRIS and UPOL, were used as primary data. From unsegmented images of UPOL database we derived other two test datasets: one using manually segmented images and another one using automatic segmented images. The distance measures defined using SIFT feature were applied for iris recognition using as input the four datasets: UBIRIS, UPOL and those two derived from UPOL. The experiments results revealed that the tested methods can be used as a simple, fast and very efficient filter, whose output can be then processed by more sophisticated methods.