Writer style from oriented edge fragments

Abstract

In this paper we evaluate the performance of edge-based directional probability distributions extracted from handwriting images as features in forensic writer identification in comparison to a number of non-angular features. We compare the performances of the features on lowercase and uppercase handwriting. In an effort to gain location-specific information, new versions of the features are computed separately on the top and bottom halves of text lines and then fused. The new features deliver significant improvements in performance. We report also on the results obtained by combining features using a voting scheme. © Springer-Verlag Berlin Heidelberg 2003.