Thinning And Target Recognition Of Fingerprint Image Under Polar Coordinates

Recognition technology using manpower instead by computer has been an inevitable trend. The traditional identity authentication system, for example, setting user name password on internet and valid documents have many disadvantages and biological recognition technology become a new main recognition technology because of high security. Fingerprint recognition, as one of biological recognition technologies occupy an increasingly important position in the field of recognition. This paper, we studied some algorithm about fingerprint. Then I improved and optimized some of these. The method of fingerprint identification under polar coordination was applied. Mainly methods in the article:First, the fingerprint image was preprocessed. The image was smoothed by normalized it. Foreground is extracted using local threshold segmentation. Gabor filter was applied to enhance fingerprint and reduce a lot of noises. The image was binaried using method of local adaptive threshold value.Second, ridge lines in local area of fingerprint image were seen as parallel arcs approximately. Then, the fingerprint was divided into blocks. We do a polar transformation on fingerprint image on blocks. Least squares was used to approach the circle arc. At last,the parallel in polar coordination were got. The fingerprint image was simplified.Third, the image was thinned by using modified KMM thinning method to delaminate the pixels. The image maintain the continuity and single pixel wide. Then to extract minutiae by using the temples that graded multilayer under polar coordination and extract singular point by the method of Poincare Index under the traditional coordination.Finally, feature matching were investigated based on point pattern, finger was classified by using singular points and their positions.