Forensic handheld toolmark examiners currently compare toolmarks (e.g. scratch marks on wire found as part of an explosive device, or on a door frame after someone broke into a house) by observing 2D images and subjectively deciding whether two marks were produced by the same source. Researchers have suggested that using algorithms could reduce the amount of human error. we present a clustering algorithm that performs very well on a dataset of 3D screwdriver marks generated in our lab under controlled conditions and a factorial design. The data allows us to cluster by screwdriver based on what force, angle, and side was used to make the mark. We hope our algorithm can be used by laboratories and help strengthen forensic science.
An Algorithm for Forensic Toolmark Comparisons

Conference/Workshop:
Joint Statistical Meetings
Joint Statistical Meetings
Published: 2022
Primary Author: Maria Cuellar
Secondary Authors: Heike Hofmann, Shane Jensen
Type: Presentation Slides
Research Area: Firearms and Toolmarks
Related Resources
An Open-Source Implementation of the CMPS Algorithm for Assessing Similarity of Bullets
In this paper, we introduce the R package cmpsR, an open-source implementation of the Congruent Matching Profile Segments (CMPS) method developed at the National Institute of Standards and Technology (NIST)…
A Pipeline for Cartridge Case Comparisons
We introduce novel sub-procedures in the pre-processing, comparison and similarity feature steps. The sub-procedures result in an improved error rate compared to our implementation of the CMC method while also…
Affect of Manufacturer Finishing Methods on Subclass Characteristics Production on Breech Faces
The following is a presentation given at Association for Firearms and Toolmarks (AFTE) (2022).
Exploration of Subclass Characteristics Manufacturing
A webinar presented during the Forensic Technology Center of Excellence (FTCoE0 Firearm and Toolmark webinar series on the progress of our CSAFE project. This study assesses the production of subclass…