| Technology of Biometric identification has been universally noticed in recent years and it is considered that this technology has a great future. Now, Automatic Fingerprint Identification is the most developed biometric technology, which has a bright future. The fingerprint recognition system mainly concludes the following steps: collection of fingerprints, pre-processing of fingerprint image, minutiae extraction and fingerprint matching.This paper has studied the matching module of Automatic Fingerprint Identification System (AFIS) using image processing and pattern recognition technique and Presents a new improved algorithm for fingerprint matching using Delaunay Triangulation which uses the idea of George Bebis for reference. Central to the proposed approach is the idea of associating a unique topological structure with the fingerprint minutiae using Delaunay Triangulation. This allows for choosing more "meaningful" minutiae groups (vector triangles), meanwhile the strategy of pre-filter is used, which not only reserves the recognition accuracy, but also significantly induces the recognition time, which solves the speed problem of large fingerprint database very well.Introduction of fingerprint recognitionCurrently, the algorithms of fingerprint matching can approximately be divided as: correlation-based algorithm, minutiae-based algorithm, ridge texture based algorithm and so on. The algorithm brought forward by this paper is based on minutiae of fingerprint.The average fingerprint matching algorithm requires the accuracy of the minutiae strictly. When we collect fingerprints, the images inevitably distort because of the unbalanced force put on the scanner, which brings about the changes of position of the minutiae and then the distance between them. And at the same time, there is angle error between the matching fingerprint and the template one. All these factors will reduce the recognition accuracy and so weaken the practicability of the system.So only if the fingerprint matching algorithm can tolerate the displacement, rotation, distortion and false minutiae, the algorithm can correctly recognition the fingerprint. Background of Delaunay TriangulationTriangulation is a process that takes a region of space and divides it into sub-regions. The space may be of any dimension. When we consider 2-D space, Triangulation deals with the limited 2-D points and connects them in a certain manner to form a group of triangles, which are not across. Generally, shapely triangles are desired.We can obtain many kinds of triangulation based a group of points, in which the Delaunay Triangles have the best quality. 2-D Delaunay Triangles satisfies the rule "the minimum internal angle is the maximum", so it is the most popular triangulation.Delaunay Triangulation has certain properties, including:the Delaunay triangulation of a non-degenerate set of points is unique;a circle through the three points of a Delaunay triangle contains no other points;the minimum angle across all the angles in all the triangles in a Delaunay if greater than the minimum angle in any other triangulation of the same points.The properties of Delaunay triangulation guarantee the following things:Once the fingerprint is given, that is a group of 2-D points are given, the Delaunay triangles are exclusive. They have nothing to do with the sequence of the points, but only with the topological structure of points. The insertion of a new point in a Delaunay triangulation affects the Delaunay triangulation only locally. That is say noise and local distortion only have local effect on Delaunay triangulation. That implies as long as there is some region in the fingerprint has not been seriously damaged, it can locate exactly and so make correct recognition possible, which is we desire;That the Delaunay triangles are not skinny is desirable in our application, since the computation of the geometric transformation is based on corresponding vector triangles. Using skinny triangle can lead to instability and errors. In...
|
PDF Full Text Request