Research On The Algorithms Of Automatic Fingerprint Identification System

Fingerprints have been used for personal identification for a long time.Fingerprints are more reliable than other biological characteristics.Although the fingerprint identification has many areas of research results, and now many businesses have developed molding products, it still has far- reaching significance.Because these core technologies can not disclose and society has set higher request.The thesis mainly includes three parts: fingerprint image preprocessing,feature extraction and fingerprint matching.In the image preprocessing, the thesis analyse the method for binarizing fingerprint image based on orientation information.The method makes holes and malposition.This thesis proposes a method for binarizing fingerprint images,which processes the original images with wiener filter. Then it calculates fingerprint direction with continuous block direction images,and enhances fingerprint images with orientation of filter templates.At last it binarizes images.The experimental results indicate that the presented algorithm achieves good capability of repairing the disconnected line,eliminating the holes and removing crossed lines in the fingerprint images.In the matching, there is deformation. The reference points of selecting is difficult.The thesis puts forward a new feature vector.It can effectively reduces the impact of fingerprint deformation. The thesis adopts two-stage matching way. It early eliminates those fingerprints whose similarity are very low .It shortens matching refused time. The two-stage matching comparison adopt adaptive threshold.It improved identification accuracy.In this thesis,experiments have been done in the PC of software MATLAB and experiment results indicate that the fingerprint verification algorithm is more effective,feasible, and it achieves the desired results.