Effects of Time and Posture Factors on Handwriting Features of Dynamic Signature

Abstract

The reliability of signature verification systems and the accuracy of forensic signature comparison are affected by factors, such as time and writing posture. Current work on posture-induced variations relies mainly on qualitative classification methods, which lack objective quantitative metrics. To understand the underlying patterns of dynamic signature handwriting under various time and postural conditions, this study collected 12 000 signature samples across five sessions, incorporating six postures. A signature posture dataset was defined using four joint angles (wrist, elbow, hip, and ankle) to quantify signing postures. Different statistical methods were applied to analyze the static and dynamic features of signature handwriting in order to evaluate the significance of time and posture influences on signature features. For the signature posture dataset, the generalized linear model and structural equation model were employed to assess the extent of joint angle effects. Principal component analysis was further utilized to identify the features that are the most significantly affected by posture. The results show that time and posture have statistically significant effects on different signature features. Posture-induced variations predominantly (80% ) arise from differences between sitting and nonsitting postures, with 69.5% of these variations attributable to changes in hip and ankle joint angles. Among the posture-sensitive features, area exhibits the highest sensitivity to postural variations. These results provide important insights for research in the fields of signature verification and forensic handwriting.