The minutiae points of fingerprint images are nothing more than point features for that image. As such, we propose an algorithm designed to detect these point features and locate these true minutiae points. This algorithm is known as the corner detection algorithm. The 'corners' that are detected are not only corners in the sense of the intersection of image lines, but also in the sense of the corner structures in patterns of intensities. Since the fingerprint image is generally thought of as a texture map with definite pattern, this approach is logical and appropriate. This thesis represents the process, comparison, and results of implementing the corner detection scheme in order to extract minutiae from fingerprint images. The outcome of the experiments will show that this algorithm is much faster than any other approach implemented thus far. This algorithm is also at least as accurate as classical methods.
展开▼
