Research On Segmentation Methods Of Fingerprint Image

Fingerprint image recognition is a means of authentication that has broadapplication prospects, and has much more advantages than any other biologicalcharacteristics. More and more fingerprint based recognition systems already usedin our daily life, such as: Work attendance, airport personnel management,evidence extraction, and so on. Fingerprint recognition already received a lot ofachievements in the area, but it is a far cry from the desired result in practice. As animportant part of pretreatment process of fingerprint image recognition, the resultof Fingerprint image segmentation will directly affect the subsequent processingresults, make great sense to improve system performance.This thesis first summarizes and compares new methods of fingerprint imagesegmentation, and then proposes an improved fingerprint image segmentationalgorithm on the research of existing segmentation algorithms, and then designs afingerprint recognition system. The main works include:(1) This thesis summarizes fingerprint image segmentation methods of recentyears. This thesis analyses and compares the methods respectively from structurecharacteristics and specific theories, and discusses the development of fingerprintsegmentation.(2) This thesis proposes a fingerprint image segmentation combining ParticleSwarm Optimization algorithm and Snake model. By improving two disadvantagesof Snake model, the algorithm solves the problems of automatic determination ofthe initial contour points and sunken areas cannot be captured.(3) This thesis designs a fingerprint image recognition system. The programintegrates the basic steps of fingerprint identification system, includes qualitycontrol module, preprocessing module, feature extraction module, matching module,and uses our algorithms in it.