Singular Points Detection in Fingerprint Images using the Partitioning Region of Orientation Fields

Abstract

Fingerprint classification is an important step in the Automated Fingerprint Identification System (AFIS). One popular approach to fingerprint classification uses singular points detection as reference points. However, it is not easy to detect singular points because it depends on the condition of the fingerprint. Fingerprints damaged by noise can result in spurious detection. This paper proposes a new algorithm for extracting the position and type of singular points in fingerprint images by combining the candidate regions of singular points (CRSP) and the Poincaré Index method. The results of the experiment using images with different image quality from NIST Special Database 14 demonstrate that the proposed algorithm has high accuracy in locating both the position and type of singular points.