Putz-Leszczyńska. J; Pacut. A
Handwritten signature verification algorithms are designed to distinguish between genuine signatures and forgeries. One of the central issues with such algorithms is the unavailability of skilled forgeries during the template creation. As a solution, we propose the idea of universal forgery features, where a global classifier is used to classify a signature as a genuine one or as a forgery, without the actual knowledge of the signature template and its owner. This classifier is trained once, during the system tuning on a group of historical data. A globalclassifier trained on a set of training signatures is not to be additionally trained after implementation; in other words,additional user enrollments have no effect on the global classifier parameters. This idea effectively solves the issue of the lack of skilled forgeries during template creation. We show that this approach can be applied both in on-line and off-line signature verification systems.