Research On The Algorithms Of Automatic Fingerprint Identification

As one of most traditional and reliable biometric technology, Automatic fingerprint identification has been applied to many fields. But fingerprint identification technology still has some unfathomed problems. Automatic fingerprint identification technology is very important for digital image process and pattern identification research not only at present but also in the future. Some of key algorithms about fingerprint image buildup and binarization are provided. The detail is as follow,(1) After studying most of filters, an algorithm based on Fourier transform band-elimination filtering is presented according to fingerprint image's characteristic. This filter can save high frequency in image and eliminate yawp in image, which greatly promoted fingerprint pretreatment efficiency. We researched fingerprint filter base on fingerprint direction.(2) Comparing some algorithms about binarization in common use, binarization based on band-elimination filter and potential difference binarization are provided, which greatly improved the retentivity of fingerprint characteristic and fingerprint binarization buildup algorithm. To meet the real needs of the database, this essay advocates a deleting arithmetic to eliminate the fringe line in fingerprint image, which effectively solved the problem and has been proven reliable in the test.Finally, according to summary of this thesis, the improvements need to be finished are analyzed, and the direction of future work is given.