Quantification of Toolmarks

To provide better customer service, NCJRS has made this Federally-funded grant final report available electronically in addition to traditional paper copies. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. characterization of toolmarks using a quantitative, statistical analysis method. Persons involved in the project include ISU faculty members L. Abstract The goal of this research was to develop a methodology to characterize toolmarks using quantitative measures of the 3-dimensional nature of the mark, rather than a two-dimensional image of the mark. Once such a methodology was developed, objective comparisons between two toolmarks could be made to determine whether marks made from similar tools could be distinguished quantitatively from marks made using other tools. The toolmarks studied were produced using 50 sequentially manufactured screwdriver tips that had not seen service. Marks were made at angles of 30, 60, and 85 degrees by a qualified toolmark examiner using a special jig. Four replicas were made of each mark, the marks being characterized using a stylus profilometer. Ten traces were taken from each mark, yielding a database of 12,000 potential data files. Initial efforts to use stereomicroscopy to obtain similar 3-dimensional information failed due to the inability to produce suitable registry between the SEM images. This problem became evident as attempts were made to " stitch " the numerous images together to obtain a continuous trace of the surface at a magnification high enough to allow quantitative measurement of the fine detail of the surface. An optical profilometer was eventually purchased that allowed suitable data to be obtained not involving touching the sample surface. The algorithm developed to allow comparison of two scans in an objective, quantitative manner mimics the procedure used by forensic examiners in that it compares the 3-d information contained in a user-specified " window " from one file to any other selected data file. The region of best fit between the two files is found, then a verification sequence is run that compares corresponding regions in the two files, selected at random, which are translated rigid, fixed distances from the region of best fit. The quality of these comparisons is evaluated using a t-statistic. If these comparisons also have a good correlation between each other a high t-statistic value is returned, indicating a high probability of a match. If the value from the rigid …