Fingerprint matching is a key issue in fingerprint recognition systems. Although there already exist many researches about fingerprint matching algorithm, it is still a challenging problem for reliable person authentication because of the complex distortions involved in the fingerprint. The non-linear deformation can lead to the change of both position and direction of minutiae and hence decrease the reliability of the minutiae. In this paper, we propose a novel minutiae-based fingerprint matching algorithm. First a new structure Adjacent Feature Union(AFU) is defined, which is invariant to rotation and translation. AFU represents the local feature of a fingerprint and it is used to align the fingerprint. Moreover, the information of the ridge frequency and block orientation is utilized to construct a distortion-tolerant model. Using this model the position and direction of the minutiae can be readjusted and thus enhance the performance of the matching. The proposed method can minimize the effect of the elastic distortion in fingerprint. Experiment results demonstrate the effectiveness of the matching method.
展开▼
