Study of the Distinctiveness of Level 2 and Level 3 Features in Fragmentary Fingerprint Comparison Krzysztof M. Kryszczuk, Patrice Morier, Andrzej Drygajlo Signal Processing Institute Swiss Federal Institute of Technology Lausanne (EPFL) {krzysztof.kryszczuk, patrice.morier, andrzej.drygajlo}@epfl.ch Abstract. In this paper we present the results of an experiment which aims to provide an insight into the problems related to the fingerprint recognition from its fragment. Level 2 and Level 3 features are considered, and their distinctive potential is estimated in respect to the considered area of a fingerprint frag- ment. We conclude that the use of level 3 features can offer at least a compara- ble recognition potential from a small area fingerprint fragment, as the level 2 features offer for fragments of larger area. 1 Introduction The studies of distinctive fingerprint features have a long tradition. As early as 1872 a probabilistic analysis of selected level two features was performed by F. Gal- ton [4]. Level one and level two features are related to the characteristic configuration of the ridges on the surface of the fingerprint. The characteristic ridge alignment in the center of the fingerprint is regarded as level one feature. The local discontinuities and links between ridges are considered level two features, otherwise referred to as the minutiae. The most prominent and common among the minutiae are the ridge termi- nations and bifurcations [7]. Level two features, are currently by far the most com- mon feature set used for automated fingerprint recognition [4]. Level three features consist of local intra-ridge details, including the traces of the sweat pores distributed over the ridges. The pores are the termination loci of the sweat ducts that originate in the sweat glands in the dermis. An early description of the pores and their characteristics can be found in [2]. The position of the pores on the ridges is thought to be individual and distinctive for a given finger. Level three features (the pores in particular) have only recently attracted the atten- tion of researchers working in the area of automated fingerprint verification [10]. The reason for that can be sought in the fact that the pores are much finer features of the fingerprint than the ridges are, and they consequently require higher image resolution