An effective method for extracting singular points in fingerprint images

Abstract In most algorithms of fingerprint identification and fingerprint classification, extracting the number and the precise location of singular points (SPs) is of great importance. In this paper, a new algorithm based on the distribution of Gaussian–Hermite moments is presented for detection of SPs. All other SPs extraction methods can only extract two types of SPs (core and delta). They often ignore a true pair of core–delta that are close to each other. Our algorithm can detect not only the core and the delta, but also a pair of core–delta. It is shown how very accurate detection of the SPs. These estimates can for instance be used for fingerprint classification and accurate registration of two fingerprints in a fingerprint verification system.