Fingerprint alignment based on local feature combined with affine geometric invariant

In this paper we introduce a novel method of fingerprint alignment that uses the intrinsic geometric properties of minutiae-based triangles combined with the geometric invariant. The minutiae points are extracted from the fingerprint image and a Delaunay (DL) triangulation is constructed from these minutiae points resulting in a series of triangles. Corresponding minutiae points are established using local affine invariants constructed from the local minutia-based triangles. Triangles that are distorted by noise or have no counter part on the query are discarded. We rely only on "strong" matches that are reliable and present, for example, where the error metric between the local absolute invariants is below a set threshold. The correspondences of such matches are then used to estimate transformation parameters. The performance of our method is represented by computing the distance map error between a template and a query fingerprint after undoing the transformation, computed from the ridge structures of the two fingerprints. In conclusion, the proposed method can be used to find the corresponding minutiae and align any fingerprints considered into affine transformation, in the presence of noise including the partial occlusion.