Abstract
With the increased use of digital devices, forensic document examiners (FDEs) encounter increasing number of dynamic or online signatures during their physical examinations. This shift expands the possibility of examinations and creates new challenges for FDEs. As such, FDEs require new examination skills using data science-based analyses with artificial intelligence and machine-learning techniques. In recent years, automated signature verification has gained significant interest in biometric research and could be useful in forensic investigations. However, the use of complex black-box systems inconveniencing FDEs in explaining the rationale behind their final assessment, especially when dealing with limited signature samples and various types of forged signatures. Therefore, a new forensic method is needed to assist FDEs’ analysis. To tackle these challenges and incorporate lessons learned from biometrics into forensics, this study proposes a novel forensic online signature analysis method. The proposed method uses a single-template strategy based on recent scientific findings in biometrics while updating the strategy for forensic use. This strategy creates a mean-template set from known signature samples that serve as a writer’s signature master pattern. Consequently, FDEs can evaluate intra-writer and inter-item variations using the mean-template set and a questioned signature. Furthermore, to take advantage of recent digital devices, we focused on both on-surface and in-air trajectories of online signatures, which could improve the discriminative power because in-air trajectories are invisible for imposters. Finally, we demonstrate the effectiveness and applicability of the proposed method in a forensic scenario using a public forensic online signature dataset.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have