The State of the Art in Fingerprint Classification

Abstract

Fingerprint classification is an effective technique that allows the number of comparisons necessary to retrieve a fingerprint in a large database to be strongly reduced: In fact, if a reliable and accurate classification is performed, an unknown fingerprint needs to be compared only to the fingerprints belonging to the same class. Automatic fingerprint classification is a very difficult pattern recognition task, due to the small interclass variability, the large intraclass variability, and the presence of noise. This chapter surveys the main approaches presented in the literature and introduces a fingerprint classification method based on a multispace generalization of the Karhunen-Love transform (MKL), which is particularly promising and achieves very good classification accuracy. Results on NIST DB4 and NIST DB14 are reported and compared with those published in the literature.