Automated comparison and evaluation of striated cutting plier toolmarks on metal wires

Toolmarks examination validity and subjectivity have come under scrutiny. This research focuses on the case of cutting plier marks. This paper presents an automatic comparison method and assesses its performance. It is designed to assign a weight to the forensic evidence (i.e, a comparison between toolmarks) with a likelihood ratio (LR). 3D topographies are acquired and treated to be compared using a set of correlation metrics. A machine learning algorithm combines comparison metrics and enables LR computation. Pliers of various brands and models were used to study the variability both within and between tools. We explained why the specific zone (area along the blade) has to be chosen to build the within-source variability and how the between-source variability can be built in different scenarios. Misleading evidence rates between 0 % and 4 % have been measured and it demonstrates the accuracy of the method when applied on the pliers used.