| Summary: | 碩士 === 靜宜大學 === 資訊管理學系研究所 === 92 === The issues of security are broadly discussed nowadays. Without effective security protection, even the most excellent information system is still useless. So far there have been a variety of biological features for personal identification, including appearance, hand geometry, voice print, iris, retina, signature, facial thermogram, hand vein, gait, odor, keystroke, and so on . Of these various methods of protection, fingerprinting is one of the most common approaches.
In this thesis, the mechanism of a novel fingerprint verification will be studied. The algorithm consists three parts, namely, the first part acquisition of fingerprint image, the acquest of features and the comparison of features. The contains the acquirement of image ridge. To detect the direction of fingerprint area, this study adopts the methods developed by Stock and Swonger to search for the whole picture of fingerprint’s flowing direction, and then present it with eight directions. The second part is based on the location of the central point. Then, the features are found according to the central points. This study uses one-class support vector machines(SVM) to distinguish fingerprints in the third part. Experiment results show that one-class SVM can apply to fingerprint recognition problem efficiently.
|
|---|
