Fingerprint Image Segmentation Method

The technology of fingerprint recognition is an active subject in the area of pattern recognition, it has broad application. The fingerprint image preprocessing is the first step of fingerprint identification system, the quality of the fingerprint image preprocessing affects directly the result of identification system.This paper investigates the theory and methods of fingerprint image segmentation—one of the important steps of the fingerprint image preprocessing. Some of the segmentation methods are discussed and classified and basing on these methods some improved methods are presented aiming at the status now in this field. The main work including:(1) Analyses some common fingerprint image segmentation methods which based on block: variance method, orient image method, compose method and the method of segment the fingerprint image in several steps. Improves the method of segment the fingerprint image in several steps, sets some of the threshold automatically ,so it can overcome some disadvantage of subjectivity and discommodiousness caused by setting threshold based on experience.(2)One of the fingerprint segmentation methods which based on pixel, the method of segment the fingerprint image based on Gabor Filter is researched and the improved method is proposed, experiment results show that the new method has the faster running speed. Improves the method of segment the fingerprint image based on Gabor Filter, enhance the speed of the algorithm. Presents the improved algorithm based on liner classifier algorithm: the method of segment the fingerprint image based on Binary Logistic regression model, the experiment results shows that this method has better segmentation results.(3) Analyses and compares some segmentation methods which based on image binaryzation: otsu segmentation method, segmentation method which based on histogram equalization and the segmentation method which proposed by S.D.Yanowitz. Analyses the qualities of these methods by experiment results.