The use of directional patterns has recently received more attention in fingerprint classification. It provides a global representation of a fingerprint, by dividing it into homogeneous orientation partitions. With this technique, the challenge in previous works has been the complexity of the pattern templates used for classification. In addition, incomplete fingerprints are often not accounted for.
Directional patterns (DPs), which are formed by grouping regions of orientation fields falling within a specific range, vary under rotation and the number of regions. For fingerprint classification schemes, this can result in misclassification due to inconsistency of patterns. Knowing the optimal angle by which to rotate the image and the optimal number of orientation regions to divide it into can be beneficial in analysing specific properties of a class. Furthermore, the number of regions directly impacts singular point (SP) detection, therefore using the optimal number of regions prevents loss of SPs. However, no previous work justifies the use of a specific number of regions or angle of rotation. More so, no explicit studies have been conducted to establish the optimal number of regions or angle of rotation that result in gaining the most information from a pattern. Therefore, this research investigates the change in DPs under the variation of rotation and number of regions to determine which condition provides the best representation of the fingerprint that is less prone to noise and minimizes interclass variability issues with fewer possible patterns for each class. This can serve as a baseline for future works using DPs. The experiments were conducted on the Fingerprint Verification Competition (FVC) 2002 database (DB) 1a. It was found that using a small number of regions produces the most accurate SPs detection and increasing the region number to more than 6 regions drastically depletes the accuracy of SP detection. Furthermore, aligning the SPs of a fingerprint containing a single loop and delta, highlights the essential properties of a class better, with fewer layouts for each class.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.