Fingerprint Registration Based On Global Information And Its Application In Matching

In today's information age, how to accurately identify personal identity and to protect information security has become a critical social problem that must be addressed. Due to the traditional identity easily counterfeit and lost and increasingly difficult to meet the needs of society, currently the most convenient and safe solution is biometric technology. It is a technique that utilizes human physiological characteristics for authentication. Fingerprint recognition technology is now one of the most widely used biological characteristic recognition technologies, which mainly involves fingerprint collection, fingerprint preprocessing, fingerprint feature extraction, fingerprint matching process. However, fingerprint matching, the core part of the automatic fingerprint recognition system, greatly affects the performance of the automatic fingerprint identification system, and is currently a hot topic in fingerprint identification technology.Fingerprint registration is one of the most critical aspects of fingerprint matching. Traditional fingerprint registration can be divided into three categories: fingerprint registration based on singular points; fingerprint registration based on minutiae-descriptor; fingerprint registration based on the Hough transform. These three types of registration are disadvantages: When there is no singular points on the fingerprint to be registered, fingerprint registration based on singular points can not be implemented; because the distinction between descriptors of minutiae is not great and the number of minutiae is not stable, it makes the accuracy of fingerprint registration based on minutiae-descriptor not high; when a large number of matching feature points exist, the efficiency of fingerprint registration based on the Hough transform is low. In order to improve the above shortcomings of the traditional fingerprint registration, this paper presents a fingerprint registration algorithm based on global information.The proposed registration algorithm based on global information can complete better fingerprint registration in spite of no singular point, less prospect area, poor fingerprint image, and can correct minutiae-based registration. The main work is as follows: First, the analysis of the current classic fingerprint registration algorithms, and the indication of their disadvantages; Second, we propose fingerprint registration based on the global information of fingerprint including orientation field, frequency field and the curvature field to complete fingerprint registration by selecting the initial registration parameters according to the principle of the steepest descent method; third, since the selection of initial registration parameters to enhance the fingerprint registration algorithm accuracy and registration time saving is very important, to select the initial registration parameters was studied; fourth, we proposed a modified minutiae matching score design strategy conducive to matching based on minutiae; fifth, the effectiveness of the proposed correlation algorithm is verified by some experiments.Fingerprint registration algorithm proposed this paper is simple, efficient and highly stable. By matching experiments on FVC2002 fingerprint database of this fingerprint registration algorithm and existing fingerprint registration algorithms, the accuracy and time complexity of the proposed fingerprint registration algorithm are better than the existing registration algorithms. At the same time, the action that the paper's registration algorithm and improved matching score are applied to Tico-based minutiae matching algorithm, reduces the equal error rate of recognition system from 1.98% to 1.01%, and significantly improves the recognition rate(on the FVC2002DB1 database, according to the standard design 2800 pairs of true match and 4950 pairs of fake match for matching experiment).