Nonparametric Fingerprint Deformation Modelling

摘要

This paper presents a novel approach to modelling fingerprint deformations. When fingerprints are captured, they undergo a certain amount of distortion due to the application of a three dimensional elastic tissue against a flat surface. This poses a challenge for automated fingerprint verification systems, which are generally based on aligning two fingerprints and comparing their respective features. There are only a few methods reported in the literature for modelling fingerprint distortions. One prominent method is based on using minutiae correspondences as landmarks, and creating a deformation model using thin-plate splines. There are several disadvantages to this approach, and a nonparametric elastic modelling algorithm is developed in this paper to address these issues. Both algorithms have been implemented and are evaluated by incorporating them into a fingerprint verification system. The results show an improvement of the proposed algorithm over the existing method of deformation modelling.