Quantitative comparison of striated toolmarks.

Area of Science:

• Forensic Science
• Computer Vision
• Pattern Recognition

Background:

• Human comparison of striated toolmarks is subjective and relies on expert experience.
• Objective and automated methods are needed for reliable toolmark analysis.
• Screwdriver toolmarks present challenges due to variations in angle and compression.

Purpose of the Study:

• To develop and evaluate an automated, objective method for comparing striated screwdriver toolmarks.
• To assess the system's robustness against variations in angle of attack and toolmark compression.
• To compare the performance of the automated system with human toolmark examiners.

Main Methods:

• Utilized multi-scale registration (alignment) to account for shift and scaling in toolmarks.
• Employed global cross-correlation as an objective similarity metric.
• Incorporated a strategy to filter spatial frequency ranges, focusing on relevant geometric details.
• Evaluated performance using 3D topography scans of experimental toolmarks from 50 screwdrivers.

Main Results:

• The automated system demonstrated high discriminatory power, accurately identifying matches even with >15° angle differences.
• The system showed greater power than human examiners in supporting common origin for toolmarks with a 30° angle difference.
• 3D data provided slightly better results than 2D data for estimating match/non-match distributions with unused screwdrivers.
• The automated method showed potential for comparing marks from other types of tools.

Conclusions:

• The developed automated method offers an objective and robust approach to striated toolmark comparison.
• The system surpasses human examiners in certain scenarios involving significant angle variations.
• Further research may extend this methodology to a broader range of toolmark analysis.