Intrinsic coordinates for fingerprints based on their longitudinal axis

Image alignment is an important step in the fingerprint recognition process. In order to obtain that Euclidean motion, its rotation and both translational components need to be determined. We propose a novel method to obtain this information by extracting the fingerprints´ longitudinal axes from their orientation fields. A model-based approach allows to additionally fix a horizontal axis, thereby defining an intrinsic coordinate system. We evaluated the robustness of this method on the NIST Special Database 4.