Iris recognition with 4 or 5 fuzzy sets

In current literature, the degrading performances of iris recognition systems is put in user’s responsibility (Biometric Menagerie -BM), or exp lained through a vague mix of t ime-related changes in biometric pattern, its presentation and the acquisition sensor (Template Ageing-TA). Actually, BM and TA are annoying symptoms illustrating intrinsic limitations of the statistical model of iris recognition. In order to avoid such limitations, a decisional model with 4/5 fuzzy sets is defined and tested here. The novelty is that among these fuzzy sets, two (the imposter and genuine score sets) are mutually exclusive and therefore, the confusion between imposter and genuine scores is no longer statistical.