Determining the similarity of signatures on the basis of characteristic points analysis

Abstract

In this study, a method of determining the similarity of signatures based on detection of characteristic points was proposed. These points were determined with the use of the IPAN99 algorithm, which detects high curvature points on a contour. A set of such points was treated as characteristic points. Up to now, the IPAN99 algorithm has not been used for finding characteristic points in images of signatures. On the basis of the obtained set of characteristic points, the similarity of signatures was calculated with the use of the ER². This study shows that reducing the number of points with the use of the IPAN99 algorithm allows obtaining a smaller error in signature recognition when compared with an analysis of all points.