Semi-automated forensic examination of handwritten character loops

This paper presents a semi-automated method for the forensic examination of handwritten character loops. The methodology is illustrated and applied to a case study designed to represent typical forensic handwriting casework, centered on a questioned acknowledgment of debt. An image analysis procedure was performed on both the questioned document and the control material provided by the person of interest (p.o.i.) to extract character loop skeletons and characterize their shape through Fourier analysis. Following this, a Bayesian probabilistic model, as proposed by Tzai et al. (2026), was utilized to assess the evidential value for discriminating between competing propositions concerning the writership of the questioned document. This specific model was developed to facilitate the joint evaluation of handwritten loops across various character types. Further experiments were conducted to evaluate the impact of data quantity on accuracy, particularly concerning misleading evidence results.