Fingerprint segmentation based on an AdaBoost classifier
Frontiers of Computer Science in China
| Hi-index | 0.00 |
Fingerprint segmentation is a crucial step for automatic fingerprint recognition. In this paper, we present an algorithm to extract fingerprint foreground using active contour model, a discriminative feature based on Fisher measure first be constructed to indicate foreground and background, however, the feature fails to describe the foreground and background when applied to those too dry fingerprints, an regulative term based on block-wise average gray level and orientation coherence be added to obtain an even distributing feature which more benefit to carry out curve evolvement. The segmenting method is testified on public fingerprint collection, FVC DB3, which contains 800 fingerprints with various image qualities. The experimental results show the effectiveness of the proposed method.