Automatic feature extraction and recognition of fingerprint images

A technique for the recognition of fingerprint images is described in this paper. Fingerprint images are preprocessed by using a set of contextual filters first and then thinned to extract the fingerprint "minutiae" features. As the images are noisy, there are many spurious minutiae. These spurious minutiae can be detected and deleted efficiently by a proposed method. The method for matching fingerprint images is also given. The usage of the technique in this paper allows for fairly high degree of noise tolerance.