Fingerprint Attributions Extraction And Their Topological Relationship Construction

The traditional identity authentication based on the password and CARDS can’t satisfythe needs of the modern people’s life and work any longer. Fingerprint identification hasmost high applicability and feasibility, which can be used in person’s identity authenticationaccording to its unique biological characteristics.There are many fingerprint identification algorithms now, but most of them ask forhigher quality of fingerprint image. The robustness of fingerprint recognition algorithm isnot high and how to improve the robustness of them is still an important and challengingresearch subject.According to the idea of topological pattern recognition, this dissertation considersfingerprint minutiae picking up, ridge tracing, choose of the attribution, topological relationconstructing based on fingerprint attribution for fingerprint recognition. The main contentsof this dissertation are here.(1) Fingerprint image is collected using fingerprint collection device and image isenhanced, binaried and thinned on the platform of Multiple documents interface (MDI)using VC++.(2) Two kinds of typical fingerprint detail feature point——the end point and the forkpoint are picked up by using3×3template and false feature points are removed on theprocessed fingerprint image.(3) A new ridge line tracing algorithm from the fork and the end point based on thethinning fingerprint is presented. All the pixel coordinates of ridge line are recorded toprepare for the ridge attribution calculating.(4) Attribution and topological relation of the minutiae and ridges are constructed toidentify the fingerprint. It is proved that this topological relation has certain invariance forthe translation, rotation, zoom and other nonlinear deformation of the fingerprint image.In the last part of this paper, the summary of the research is given and the researchdirection in future is also pointed out.