Fingerprint image postprocessing: A combined statistical and structural approach

Abstract For complex patterns such as fingerprints, it is difficult to describe the dactylotype by using a set of feature vectors, as well as to construct and parse all of the individual fingerprint pattern with a syntactical approach. In this paper, which is different from the conventional methods, we present an approach in which the statistical and structural approaches are combined to represent the fingerprint false minutia patterns. Based on the proposed approach, a fingerprint image postprocessing algorithm is developed to eliminate the false minutiae in fingerprint images. Because a global minutia relation is considered on the whole image, experimental results demonstrate the advantages of the combined approach.