Research Of Fingerprint Identification Algorithm Based On Multiple Reference Points And The Principle Of Similar Triangles

With the rapid development of electronics and communications technology,more and more high technology is applied to people's daily lives.Coupled with the popularity of the network and continuously update of our technology,people's attention on information security increases.Traditional identification technology has far failed to meet the high demands of today social security.The rapid development of biometric technology can solve these problems fundamentally.Especially,the unique,universal,easy collection,high anti-counterfeiting features of the fingerprint identification technology are called the most promising biometrics.The fingerprint identification system generally consists of four main components:acquisition of a fingerprint image,preprocessing of a fingerprint image,minutiae feature extraction of a fingerprint image,feature matching of a fingerprint image.This paper summarizes the current development,advantages and principles of fingerprint identification technology,and carries out on this subject some research. The main work of paper around the following sectionsFirstly,the fingerprint image preprocessing,traditional methods normalize the fingerprint image;the paper focuses on image enhancement section,where the fingerprint image enhancement process in the direction of the field in the strike,the paper proposes a method the gradient field based on the direction of the point the improved algorithm, every two sampling method for compressing fingerprint images,significantly reducing the time to strike a point direction of the field. And the paper is based on a recursive Gaussian Gabor filter enhancement to get a good figure;Then the paper binarizes image by combining fingerprint image orientation field binarization.The last the paper uses an layer by layer iterative algorithm to thin the binarized fingerprint image,in order to get only one pixel width and clear fingerprint image.Secondly,the fingerprint image minutiae feature extraction,the paper uses based refining image feature extraction algorithm,extraction of specific steps:Extract category and location coordinates of the point of view of a pixel based on 8 neighborhood points,then 8 neighborhood coding feature tracking algorithm to extract minutiae direction,then build to the scale,which contains the coordinates, directions and coordinates. The final feature vectors obtained minutiae are stored in the profile template.Thirdly,the fingerprint image feature matching,minutiae-based matching algorithm is a reference point,this paper once a match based on multiple reference points,and on this basis,not correctly be identified and detailed identification is rejected secondary point matching based on the principle of similar triangles.In order to demonstrate the feasibility of the proposed algorithm,FVC2004 fingerprint database samples are simulated on MATLAB R2012 a.According to the experimental results,that the combination of two algorithms greatly reduce the rejection rate and error rate,and improve the automated fingerprint identification system reliability for the future development of fingerprint recognition technology provides a certain reference value.