The research and improvement of fingerprint matching algorithm

Abstract

For the superiority of good uniqueness, stability and being easily collected, fingerprint has become one of the most widely used biological characteristics. The algorithm of fingerprints identification is generally divided into three parts: image pretreatment, feature point extraction and matching. This paper analyses and improves the most widely used matching algorithm-the shortest distance point matching algorithm which is based on point model. Firstly, we need to find a reference point, which is taken as the Origin of the polar coordinates, among the feature points. Secondly, transfer the feature points to the polar coordinates from the rectangular coordinate system, now the consequence caused by the translation of the fingerprint image is clear. There are two steps in matching process, in first matching, if the amount of the matching points, without considering the effect of angle, is greater than or equal to the threshold, we step into the second matching based on angle. It is proved that the improvement is significantly effective to the system which includes a large amount of template images in the database.