Quantitative Algorithm for the Digital Comparison of Torn Duct

• NCJ Number: 250661
• Author(s): William Ristenpart; Frederic Tulleners; Alicia Alfter
• Date Published: March 2017
• Length: 55 pages
• Grant Number(s): 2013-R2-CX-K009
• Agencies: NIJ-Sponsored
• Publication Type: Research (Applied/Empirical); Report (Study/Research); Report (Grant Sponsored); Program/Project Description; Grants and Funding
• Award: Quantitative Algorithm for the Digital Comparison of Torn Duct Tape

Annotation

The goal of this research was to minimize human contextual bias in decisions about whether torn duct tape found at a crime scene matches a duct-tape roll found in a suspect's possession by combining digital image analysis and an objective, quantitative algorithm in assessing the likelihood of a match.

Abstract

Researchers performed edge detection and morphological smoothing operations on high-resolution images (1200 dpi) of torn duct tape edges to extract the torn-edge coordinates. The coordinates of a given exemplar and a suspect sample tear were then compared by calculating the sum of square residuals (SSR) of the two sets of coordinates, producing a single quantitative number that represented the "closeness" of the match. The analysis of 11 cohorts of 200 torn pairs yielded 2,200 total pairs with 440,000 quantitative inter-comparisons, showing that SSR values on the order of or less than 105 mm2 have high probability of being a match. In 97 percent of all examined tears, the true match had the lowest observed SSR, with false positive rates ranging from 0.5 percent for some types of hand-torn duct tape to 62 percent for scissors-cut duct tape. This work provides a starting point for quantitative assessment of the likelihood of physical end matching of duct tape without human contextual bias. 13 figures, 4 tables, and 9 references