Font Size: a A A

Research On Pre-processing Technology Of Fingerprint Image

Posted on:2006-08-04Degree:MasterType:Thesis
Country:ChinaCandidate:S B HuFull Text:PDF
GTID:2178360185463326Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
It is very important for minutia extraction in AFIT (Automatic Fingerprint Identification Technology) based on minutia matching. The fingerprint image always has much noise and can not be adapted to extract minutia directly, so pre-processing of fingerprint image to get a point-line image is an important part of AFIT domain.The paper introduces briefly the characteristics of AFIT, the significance of the fingerprint image pre-processing and all parts of AFIS (Automatic Fingerprint Identification System), then expatiates on the image pre-processing included image segmentation, image enhancement, image binarization and thinning.In image segmentation, the paper improves some common algorithm of gradual segmentation according to its shortcomings. In the first segmentation step, the paper separates one fingerprint image to four parts at first, then confirms foreground board regarded as seeds by gray-level statistical features and apply region growing method, at last combines the four parts to one image that is the first segmentation image. In the second segmentation step, the paper signs the irrecoverable region from the first segmentation image by gray-level and directional statistical features. The improvements make segmentation more effective.In image enhancement, the paper improves the algorithm of fingerprint image enhancement based on Gabor filter according to its disadvantages. At first, the paper confirms singular area by singular detection, then designs a series of low-pass filters that have different templates according to the different position in finger image for a local area, at last sets up Gabor filter parameters according to the characteristics of fingerprint image. The improved enhancement algorithm has a better robust, and can preserve texture structure of singular area effectively.In binary noise clearance, the paper presents an algorithm that clears the biggish white speckles and black speckles. The algorithm has an effective clearance for the noises.
Keywords/Search Tags:AFIS, Directional image, Segmentation, Enhancement, Binarization, Thinning
PDF Full Text Request
Related items