Research And Application Of Automatic Fingerprint Recognition Algorithm Based On Point Pattern

This paper discusses the AFIS from fingerprint classification, fingerprint core automatic seek and fingerprint minutiae matching. In every process, it selects two arithmetics to make experimentations. In addition, it improves some arithmetics. Base on these arithmetics , it designs and achieves an identity verification system of depositors.These are the primary contents of this paper:First, we adopt the block direction filtering method, the self-adapt two-valued disposing method and moulding board mining method to accomplish the pretreatment of the original fingerprint image.Second, we adopt Poincare Index method and local direction searching method to seek the core of the fingerprint. In local direction searching method, it adopts thick search and thin search. These two steps realize accurate orientation of the fingerprint core.Third, using singular points and direction image, we class the fingerprint. We dispart four small fields in the center field of the fingerprint image. Through the directions of the lines in every small field, we confirm the fingerprint type.Fourth, we count the cross number of every pixel in the fingerprint images which have been thinned. Through the cross number, we confirm the minutiae feature.Fifth, we select two methods of point pattern to match minutiae feature. In the match method based on picture emendation, we use the changeable sized boundary box which influences the disadvantage of the unchangeable sized boundary box. In local vector match method, we add the line numbers and the dispersion of direction angle between the center point and near minutiae point. We also add the direction angle of the center point. These additions enhance veracity of the minutiae featurematching.Sixth, It designs and achieves an identity verification system of depositors.