Fingerprint verification using characteristic vector based on planar graphics

This paper describes a new characteristic vector model for fingerprint representation that uses planar graph and triangulation algorithms. It is shown that this new characteristic vector model presents a better performance in a fingerprint identification system when compared with other vector models already proposed in literature. The minutiae extraction is an essential step of a fingerprint recognition system. In this paper is also presented a new method for minutiae extraction that explores the duality ridge ending/ridge bifurcation that exists when the skeleton image of a fingerprint is inverted. It is shown that this new extraction method simplifies the computational complexity of a fingerprint identification system.