Research Of Fingerprint Recognition Algorithm Based On Large-Scale Fingerprint Database

Biology identify technique bases on the physiological or behavior characteristics, it identifies ID more safty and shows more convenience in study and use.Fingerprint recognition technology is the pioneer technique in biology identify field, it has the most veracity and conveniency,so that it is used in more and more domains.The aim of this paper is to design a fingerprint recognition arithmetic which is based on the large-scale fingerprint database. After satisfing the requirement of reliability, the small fingerprint recognition system must realize the fast recognition and have more convinece to human. The arithmetic must have low complication, high operation rate, so it can be content with real time demands.The shortcoming of the existing fingerprint recognition arithmetic has been discussed. Based on the characteristic of large-scale fingerprint database in matching time, some improvements to the existing previous process arithmetic, pick-up minutia arithmetic, fingerprint matching arithmetic have been maked.Then the arithmetic has been carried out in Matlab. Through experiment, its reliability and efficiency has been validated.Some aspects of work about automatic fingerprint recognition arithmetic have been done:1.The exiting elimination algorithm of false features in the choosing of threshold has been improved. The different threshold has been choosed basing on different person. Automatic threshold has been used before the elimination false features.It has provided a reliable criterion for the elimination algorithm of false features.2.The method of fingerprint classification has been analyzed. The shortage of the previous classification method which bases on singular points has been pointed out. A new classified mothed which integrates the singular points and the ridges of the fingerprint has been created to classify the fingerprint.3.The fingerprint has been matched by two steps. To abtain high operation rate, in the first step, a ROI in the fingerprint has been marked and then the number of ending and bifurcation in ROI has been calculated separately to achieve cursory match. In the second step, the changeable sized boundary box which have more robust on non-linear transfigure has been used to realize the accurate match.4.The information of the fingerprint ridges has been made good use to locate the reference point. By calculating the minimum sum of the two parameters, reference point has been lacated fast and accurately.Advanced theory and technology for the application of automatic fingerprint recognition in commercial field has been provided in this paper.