Advanced Scan Matching, Scalable Search, and Visualization Tools for the Analysis of 3D Scans of Cartridge Casings in Firearm Forensics

As submitted by the proposer:
The proposed work includes critical next steps for developing and deploying a low-cost, fast, and accurate next-generation system for toolmark analysis. In 2013, we began development of a 3D surface topography imaging and analysis system for firearm forensics. To date, we developed prototype software and hardware, conducted a series of deployment studies with local and state forensic labs, completed matching experiments on a large dataset of several thousand casings, devised matching algorithms for breech-face impressions and aperture shears, evaluated cross-modality matching (i.e., the matching of a GelSight scan with a confocal scan), developed scanning performance metrics, established best scanning practices, and established a quality control process. Overall, the technology shows excellent (and improving) match accuracy and has the potential to greatly impact the analysis of toolmarks in the laboratory. This project has been a collaboration with local, state, and federal crime labs from the beginning. The in-field testing provided by these labs has strongly influenced project development. This year (2016) we are extending our imaging and analysis methods to include firing pin impressions, developing and evaluating a blind verification tool, and researching virtual microscopy.

Building from our recent success in the scanning and analysis of cartridge casings, the proposed work (2017) takes important next steps towards moving the developed technology into the crime lab. The proposed work develops a recall-focused scoring function capable of identifying matches for minimally marked casings (a common occurrence in the lab), develops a distributed computing search algorithm which will greatly reduce both the time and cost of conducting database searches, and develops a free 3D casing viewer for the entire forensics community based on the newly established X3P 3D file standard. Through all aims we will work with our forensic and academic collaborators and continue deployment studies with several sites.
These goals represent a series of R&D steps towards the creation and validation of a novel technology platform for comparing 3D surface topographies for firearm forensics. The assembled project team and committed colleagues include PhD-level computer scientists, forensic firearms examiners, and metrology experts each strongly committed to getting the research right and advancing the field. Overall, this work will develop analytic techniques, grounded in mathematical science and able to provide accurate quantitative sample comparison and database search. This should benefit law enforcement and their ability to present forensic evidence in the courtroom.

Note: This project contains a research and/or development component, as defined in applicable law.

ca/ncf