Research On Key Technologies In Automatic Fingerprint Identification System

Automatic fingerprint recognition technology is a biometric identity authentication method widely used in people's lives,such as door locks,bank ATMs5 mobile phones,safes,etc.Although many breakthrough research results have been obtained in the field of fingerprint identification,there are still problems in the automatic fingerprint identification system,and with the continuous development of information technology,people put forward higher requirements for fingerprint recognition technology.Therefore,in this paper the overall performance is improved through the related research on fingerprint preprocessing,fingerprint retrieval and fingerprint matching.The specific research contents are as follows:Firstly,the preprocessing process of fingerprint images is studied,including fingerprint segmentation,orientation field estimation,enhancement,binarization and thinning,and the specific algorithm implementation of each part is given.In this paper,the algorithm of orientation field correction is proposed for low-quality fingerprint images such as flaws and blurs.On the basis of the orientation field estimation using the gradient method,the correctness of the orientation block is discriminated according to the characteristics of the slowly changing ridge,and the corresponding method is used to correct the error orientation block,which effectively improves the accuracy of the orientation field estimation.Theoretical analysis and experimental results show that the fingerprint preprocessing algorithm has better repair effect on low quality images such as flaws and blurs,and ensures the accuracy of feature extraction.Secondly,with the increasing scale of fingerprint database,fingerprint recognition efficiency becomes crucial.Aiming at this kind of problem,a fingerprint indexing algorithm based on minutiae descriptor is proposed in this paper.First,the minutiae descriptor is used to filter the minutiae points,then the best reference point is deteraiined according to the global positional relationship between the minutiae points,and finally the similarity of the two fingerprint images is calculated based on the best reference point.By using the Search strategy through threshold method,the template fingerprint of which the similarity value satisfies the condition directly enters the fine matching,which effectively solves the defect of the Top N method.Theoretical analysis and experimental results show that the proposed algorithm outperforms the indexing algorithm based on geometric features between minutiae points in time and space,and the robustness of the algorithm is better.Thirdly,for the problem of low-quality fingerprint matching success rate such as small area,low quality and severe deforaiation,a fingerprint matching algorithm that combines minutiae and fine line structure is proposed in this paper.Firstly,the feature point information is used to find the best reference point quickly and accurately;then two fingerprints are aligned to perform the detail point matching and the fine line structure matching respectively.Finally,according to the Bayesian theory of the minimum error rate,the addition criterion is used to fuse these two feature,to determine whether the fingerprint matches.Theoretical analysis and experimental results show that the fusion matching algorithm makes up for the traditional problem of single-point pattern matching,and effectively improves the accuracy of fingerprint recognition.