A Novel Multi-reference Points Fingerprint Matching Method

Keming Mao, Guoren Wang, Changyong Yu, Yan Jin

Abstract: Fingerprint matching is a challenging problem due to complex distortion in fingerprint image. In this paper, a multi-reference points matching method is proposed to solve the problem. First, a new feature description Minutiae-Cell, which is constructed by the minutiae and its neighbor ridges, is used to represent the local structure of the fingerprint. The proposed matching method consists of three stages, including the original matching stage ,the purifying stage and the fingal matching stage. In the original matching stage, minutiae pairs that potentially matched are found based on the Minutiae-Cell and the. Then the purifying stage is carried out to obtain the true matched minutiae pairs. Instead of using only one reference pair, the final matching stage deals with the remaining minutiae from template and query fingerprints by comparing their distance to the true matched minutiae pair set. The matching score is composed of the results of purifying stage and final matching stage. The proposed method overcomes the problems of distortion and noises existing in the fingerprint image. Experimental results show that the performance of the proposed algorithm is satisfying.