Publication

Man vs. Machine: A Comparative Analysis for Signature Verification

Muhammad Imran Malik, Marcus Liwicki, Andreas Dengel, Bryan Found

In: Journal of Forensic Document Examination (JFDE) 24 Pages 21-35 Association of Forensic Document Examiners (AFDE.org) 12/2014.

Abstract

Signatures have been used as a means to authenticate documents for centuries. From the outset, the focus of forensic examinations was to both objectively and subjectively establish whether they were genuine (written by the specimen author) or simulated (written by an imposter/forger). With the emergence of new computing technologies, additional objective examination techniques designed to determine the authenticity of questioned signatures became available. Although the opinions of Forensic Handwriting Examiners (FHEs) remain the most popular method of signature authenticity determinations, computer based techniques are attracting increasing interest within the forensic community. The question here is; which is better: man or machine? To address this question we focus on empirically comparing the performance of the two, on the same or similar material. The novelty of this work is that we have applied various state-of-the-art signature verification systems to questioned signature problems which had already been worked by FHEs and then performed a comparative analysis of the two.

Weitere Links

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz