The ability to uniquely associate a human fingerprint with an individual has become an essential part of personal identification for security purposes and law enforcement. This article presents an approach to fingerprint recognition based on circularly sampling the fingerprint around an easily identified center of the fingerprint. Three matching metrics are presented and compared for matching fingerprints with no peculiarities, a missing area, and rotation with those in a database. The optimum metric, referred to as the angular density, was seven times better for a self match than for a mismatch, even when the fingerprint was rotated or altered by erasing some of its area.