Research And Realization Of The Fingerprint Identification System

Automated Fingerprint Identification System(AFIS) is started from 1960's, it has replaced conventional method to identify fingerprints with computers. Due to the uniqueness and invariability of fingerprints, the automated fingerprint identification is becoming an attractive alternative to the traditional identification methods. These years, AFIS is applied more and more aspects in our daily life with the development of computer technology, the appearance of cheap fingerprint sensors and effective algorithms. However,, there is much special noise in fingerprint images, and many methods wouldn't be publicized, so an ideal automated fingerprint identification system(AFIS) is still a difficult research object.This dissertation focuses on Automated Fingerprint Identification(AFIS), Attention has been given to Parts algorithms of the preprocessing,minutiae extraction and matching of AFIS based on minutiae for further improvement. The main contribution of this thesis includes:Two methods of segment fingerprint from background are studied: variance method and direction method. In addition, we have been improved the variance method. Fingerprint enhancement is processed through Gabor filter, utilizes average gradient method based on Sobel operator to compute point directional image. in the binarization, Binarize the gray-scale image with local self-adaptive binarization smoothness algorithm and eliminate the noises from the binary image with fast Fourier transform algorithm. In the processing of the fingerprint thinning, use mathematical morphology method to thin image.In the aspect of feature extraction, a fast and effective feature extraction and post-processing algorithm is selected.In the matching section, choosing a fingerprint-matching algorithm based on the vector triangle, This method utilizes information of two nearest minutiae to determine the reference points in fingerprint image and to fulfill fingerprint matching. The algorithm is realized,debugged and tested successfully with C++ Builder on PC, and the experimental result shows that the algorithms almost meet the basic requirement of fingerprints identification.