The Template Selection in Biometric Systems Based on Binary Iris Codes

Since the variability of data within readings from the same person is intrinsic property of every biometric system, the problem of finding a good representative - the template - was recognized and present since the beginning of biometrics. This problem was solved differently for different biometric types, yet usually the template somehow averages the collected data samples. However, for the iris type, the template is usually just one or a few samples. In this paper we describe the experiments that suggest that the averaging is also justified in case of iris template creation. This is an important fact, which can significantly improve a performance of biometric template protection methods for iris.