In spite of the numerous algorithms for automating the verification of handwritten signatures that have been proposed over the years, there are still many handwritten signatures that can cause a system to underperform significantly. Two main elements are responsible for this: the variability of a signature and its complexity. Studies reported analyse these elements in much detail, and include an estimation of a signature's complexity in terms of the difficulty that a potential forger would encounter in an attempt to reproduce it dynamically. This paper introduces a technique to estimate the complexity of a signature image based on knowledge acquired from a human experiment, where 'complexity' is viewed from a different perspective. Importantly, the measure we propose is based entirely on the static signature image.
|Number of pages
|Published - 2002