Signature Verification for Offline Skilled Forgeries Using Textural Features

Abstract

This study explores the effectiveness of two texturalmeasurements on signature verification for skilled forgeries. These texture features include 2D autoregressive coefficients andrun-length distributions. Signature images corresponding to 521writers from the GPDS960 database were used to evaluate theperformance of these features. Comparison of the proposedtextural features with a number of state-of-the-art featuresrealized interesting results. The run-length features outperformother features for a sufficient number of genuine signatures inthe training dataset.