Study On Improved Fingerprint Identification Algorithm Based On Delaunay Triangulation

Fingerprint identification is one of the most reliable personal identification methods.An automatic fingerprint identification system (AFIS) is widely employed. It plays a vital role in providing forensic evidences as well as in our everyday lives, for instance, in network safety, criminal identification, access control, and ATM card verification. Thus, to execute deep researches on fingerprint recognition is not only theoretically but also practically meaningful. In recent years, automatic fingerprint identification technique has been a research focus in the area of science. Many researchers have done much work on fingerprint identification.The matching of fingerprint which is the most important part in fingerprint identification technique has drawn more attention from researchers.This paper has studied the preprocessing of fingerprint image, minutiae extraction and the image matching using the idea of George Bebis for reference. Also Delaunay Triangulation is applied to the matching of fingerprint image.The main contents are summarized as follows:Firstly, this paper has studied many issues on the preprocessing of fingerprint image including image division, enhanced image, binarization and image fragmentation. The presently popular algorithm in the preprocessing of fingerprint image is introduced and then related theories and technologies adopted.Secondly, this paper studies on the minutiae extraction of fingerprint and improved the algorithm using 8-neighbour coding ridge tracing in the actual test.Especially the proposed approach in this paper is not only based on the previous matching algorithm but also an improved method in minutia extraction recognition with the Delaunay Triangulation. This is the idea of associating a unique topological structure with the fingerprint minutiae which is much better than the original algorithm. This allows for choosing more "meaningful" information on fingerprint which was ignored before, meanwhile enlarging dimensions of matching minutiae, which not only improves the matching accuracy, but also significantly induces the recognition time by filtering much false matching information.In addition, many tests have been done for the improved algorithm in this paper, including the evaluation on parameters, the distribution of matching scores, ROC curve and matching time etc. Also the comparative testing between the original and the improved algorithm has been performed. The results from all testing have proved that the algorithm in this paper is a bigger improvement on recognition rate or speed of matching.