A Multi-Stage Fingerprint Image Segmentation Method
Segmentation is an important step in fingerprintrecognition ystem,in which the region of interest canbe extracted,his paper presents a novel fingerprintimage segmentation method aiming at dealing withlow quality fingerprint images.Firstly the mean andvariance of gray level in each divided block is used todifferentiate the image coarsely.Then two newfeatures intra-consistency and extra-consistency aredefined in further segmentation stage and validationstage respectively.They can reflect the texturefeature within a block as well as between neighborblocks,and are utilized to partition the image refinedly,especially for the shadow area and the boundarybetween foreground and background.Gabor filter isused in this paper and a fast implementation of Gaborfilter is adopted Experimental results show theeffectiveness and robustness of the proposed method.
Keming Mao Guoren Wang Changyong Yan Jin
Institute of Computer System,Northeastern University
国际会议
厦门
英文
1141-1145
2008-11-17（万方平台首次上网日期，不代表论文的发表时间）