On Feature Extraction using Region-based Measures for Fingerprint Recognition

Most of the Automated Fingerprint Identification Systems (AFIS) are based on the minutiae (ridge endings and bifurcations) matching. In low quality fingerprints, the automatic detection of the minutiae is a very critical process. Further minutiae based matching has difficulty in quickly matching two fingerprint images containing different number of unregistered minutiae points. Real-time applications require the results with minimum false positives and true negatives [1]. This essentially demands more features to be extracted which will minimize the error in recognition. In this paper, we have designed a feature extraction strategy to extract desired (user defined) number of features [2]. This paper describes a simple fingerprint recognition system that constitutes fingerprint preprocessing, feature extraction and matching. The proposed method uses region-based measures for feature extraction. The technique uses very less number of minutiae details and tries to gather the information from every region of the fingerprint. Finally, the extracted features of the query fingerprint are matched with the template database to recognize the fingerprint