Fingerprint Matching Algorithm Based on Voronoi Diagram

In this paper we present a new indexing-based approach to fingerprint matching technique. Our proposed approach is the idea of associating a unique topological structure with the fingerprint minutiae using Voronoi diagram (VD). We utilize well-known algorithms and techniques from the initial step of image registration up to feature extraction (Minutiae extraction). Several image processing techniques such as orientation, segmentation, Gabor Filter, binarization, and thinning are employed in order to have a valid feature extraction. Using VD on the topological structure of minutiae set, results in a VD with minutiae as vertices. In the VD diagram we find a central cell; and such a region is unique based on our approach. The central cell (CC) is used for local matching. This approach rejects non-similar fingerprints instantly and has good accuracy and performance in comparison with existing algorithms. In addition, we also propose an alternative algorithm that is customized in order to take into account, distortion (non-linear transformation) based on our primary algorithm. The experiment is performed on FVC 2006 databases, and the result indicates the validity of our algorithm.