Skip to Main Content
Fingerprints are the most used biometric feature for person identification and verification in the field of biometric identification. We put forward a new fingerprint identification algorithm that combines point match and image match. This algorithm depends on singular point abstraction and wavelet transform coefficient. First singular points are extracted for rough match and image calibration. Because the fingerprint image after wavelet transformation obey Generalized Gaussian Distribution approximately, we can use two parameters to represent for wavelet sub-band characteristics, and it will greatly reduce the number of feature point and the time for match. The result of final experiment in which we used FVC2002 fingerprint database shows that this algorithm can effectively overcome the influence of image rotation and make the final recognition rate reach 90.5%, the average speed of recognition is about 0.9s.