Off-line Signature Verification Using Flexible Grid Features and Classifier Fusion

In this paper we present two novel off-line signature verification systems, constructed by combining an ensemble of eight base classifiers. Both score-based and decision-based fusion strategies are investigated. Each base classifier utilises the novel flexible grid-based feature extraction technique proposed in this paper. We show that the flexible grid-based approach consistently outperforms the existing rigid grid-based approach. We also show that the combined classifiers outperform the most proficient base classifier. When evaluated on Dolfing´s data set, a signature database containing 1530 genuine signatures and 3000 amateur skilled forgeries, we show that the combined classifiers presented in this paper outperform existing systems that were also evaluated on this data set.