Manipulative Virtual Tools for Tool Mark Characterization
S. Zhang, L. S. Chumbley
See also the instructional videos on "Impression Evidence"
Abstract
The purpose of this research is to provide quantitative data analysis tools for characterization of tool marked surfaces to improve the scientific basis of toolmark identifications. The methodology used will be an integrated approach that combines surface characterization using an optical profilometer to enable development of advanced visualization techniques to analyze and simulate striated surfaces. Specifically, thisresearch was to develop a methodology whereby a three-dimensional (3-D) computer simulation of a tool tip is generated. This “virtual tool” can then be used to produce “virtual toolmarks” - a series of predicted markings where the applied force, twist of the tool, and angle of attack of the tool tip can be varied. Quantitative 3-D data from the suspected tool and evidence toolmarkwas acquired and a virtual reality program developed that takes this data and reconstructs a “virtual tool”for computer manipulation to create “virtual toolmarks”. Since the “virtual tool” can be manipulated to produce a range of markings, the exact parameters required to obtain the best possible match to the actual tool mark can be found. Duplicate marks based on these results can then be statistically compared and ranked on the basis of quantitative measurements.
The developed software includes four major parts: (1) automated noise reduction for 3D data of both tools and real marks directly coming from an optical profilometer; (2) arbitrary virtual toolmark generation for any given tool; (3) easy to use software graphical user interface (GUI) for data visualization, manipulation, interaction; and (4) integrated computer-based objective comparison algorithms to provide statistical measure of a pair of toolmarks.
Initial experiments were based on statistically analyzing 6 different tools sampled from the 50 sequentially manufactured screwdriver tips, and 34 actual toolmarks made by a qualified toolmark examiner using a special jig. These scans were carefully cleaned to remove noise from the data acquisition process and assigned a coordinate system that mathematically defines angles and twists in a natural way. Using the virtual tookmark generation method, virtual marks were made at increments of 5 degrees and compared to a scan of the real tool mark. The previously developed statistical algorithm performs the comparison, comparing the similarity of the geometry of both marks to the similarity that would occur due to random chance. Finally, the method informs the forensics examiner of the angle(s) of the best matching virtual mark, allowing the examiner to focus his/her mark analysis on a smaller range of angles and twists.
Experimental results were very promising. In a preliminary study with both sides of 6 tips (and 6 x 2 x 13 = 156 virtual marks) and 34 real marks, the method can distinguish matches from non- matches with only a few false negatives reported (i.e. matches mistaken for non-matches). For matches, it can also generally distinguish between marks made at high and low angles with good prediction.The experimental data indicated that angle for the real mark predicted by the virtual mark could often be achieved to within five degrees of the actual angle.
In summary, the question posed as the goal of this study, “can a manipulative “virtual” tool be made to generate “virtual marks” for quantitative and objective toolmark characterization?” has been answered in the affirmative given the right conditions. Factors affecting the correct identification include the quality of the marking, suitable noise cleaning techniques, suitable virtual mark making approach, and the suitable statistical routine. Moreover, this method presents a unique opportunity to improve tool mark analysis by saving examiners’ time and reducing the possible damage to evidence.