A Preliminary Study of the Usage of Similarity Measures to Detect Singular Points in Fingerprint Images

One of the most prominent features of a fingerprint is the presence of singular points, which are locations of the fingertip at which unusual ridge patterns take place. They allow the classification of fingerprint images into different subclasses, accelerating the posterior matching process, but they can also be used to improve matching accuracy. In this work, we put forward a new method for singular point detection based on similarity measures. These measures are used to compare the orientations in a fingerprint orientation map to those in pre-established templates representing the canonical form of different singular points. This method provides a simple, yet effective, way to detect singular points in fingerprint images. Moreover, it is more flexible than the commonly used Poincare method and also significantly simpler than other approaches, such as those based on complex filters. Preliminary experiments on two datasets show promising results.