A Fingerprint Retrieval System Based on Level-1 and Level-2 Features

This paper proposes a novel fingerprint retrieval system that combines level-1 (local orientation and frequencies) and level-2 (minutiae) features. Various score- and rank-level fusion strategies and a novel hybrid fusion approach are evaluated. Extensive experiments are carried out on six public databases and a systematic comparison is made with eighteen retrieval methods and seventeen exclusive classification techniques published in the literature. The novel approach achieves impressive results: its retrieval accuracy is definitely higher than competing state-of-the-art methods, with error rates that in some cases are even one or two orders of magnitude smaller.