A New Model of Fingerprint Retrieval Based on Features of Minutiae and Gabor

Fingerprint is the commonly used biometric property in security, commerce and forensic application. One common problem in pattern recognition is lack of samples, only a few fingerprint samples from each individual are available for training a classifier. This paper proposes an approach of fingerprint retrieval based on Bayes classifier by combining the features of Gabor and Minutia and attempted to tackle the problem of insufficient training samples by generating additional samples using spatial modeling. With the expanded training set, we are then able to employ a more sophisticated classifier such as a Bayes classifier for recognition. We apply the proposed method to build a fingerprint retrieval system that is accurate and efficient. In fingerprint indexing / retrieval, the problem of one-toone matching is extended to one-to-N matching, the system searches through the entire database (FVC, NIST-4) of enrolled templates and returns a list of probable fingers (identifiers of individual) that the fingerprint may belong to. The accuracy and speed are evaluated using FVC database and the system performs better than that of KNN classifier has the drawbacks of being comparatively slow and less accurate. Keywords—Fingerprint Indexing, FVC, NIST-4, Spatial modeling, Fingerprint Retrieval.