Virtual and simulated striated toolmarks for forensic applications.

Large numbers of experimental toolmarks of screwdrivers are often required in casework of toolmark examiners and in research environments alike, to be able to recover the angle of attack of a crime scene mark and to determine statistically meaningful properties of toolmarks respectively. However, in practice the number of marks is limited by the time needed to create them. In this article, we present an approach to predict how a striated mark of a particular tool would look like, using 3D surface datasets of screwdrivers. We compare these virtual toolmarks qualitatively and quantitatively with real experimental marks in wax and show that they are very similar. In addition we study toolmark similarity, dependent on the angle of attack, with a very high angular resolution of 1°. The results show that for the tested type of screwdriver, our toolmark comparison framework yields known match similarity scores that are above the mean known non-match similarity scores, even for known match differences in angle of attack of up to 40°. In addition we demonstrate an approach to automatically recover the angle of attack of an experimental toolmark and experiments yield high accuracy and precision of 0.618 ± 4.179°. Furthermore, we present a strategy to study the structural elements of striated toolmarks using wavelet analysis, and show how to use the results to simulate realistic toolmarks.

Addressing the National Academy of Sciences' challenge: a method for statistical pattern comparison of striated tool marks.

In February 2009, the National Academy of Sciences published a report entitled "Strengthening Forensic Science in the United States: A Path Forward." The report notes research studies must be performed to "…understand the reliability and repeatability…" of comparison methods commonly used in forensic science. Numerical classification methods have the ability to assign objective quantitative measures to these words. In this study, reproducible sets of ideal striation patterns were made with nine slotted screwdrivers, encoded into high-dimensional feature vectors, and subjected to multiple statistical pattern recognition methods. The specific methods employed were chosen because of their long peer-reviewed track records, widespread successful use for both industry and academic applications, rely on few assumptions on the data's underlying distribution, can be accompanied by standard confidence levels, and are falsifiable. For PLS-DA, correct classification rates of 97% or higher were achieved by retaining only eight dimensions (8D) of data. PCA-SVM required even fewer dimensions, 4D, for the same level of performance. Finally, for the first time in forensic science, it is shown how to use conformal prediction theory to compute identifications of striation patterns at a given level of confidence.

Forensic surface metrology: tool mark evidence.

Over the last several decades, forensic examiners of impression evidence have come under scrutiny in the courtroom due to analysis methods that rely heavily on subjective morphological comparisons. Currently, there is no universally accepted system that generates numerical data to independently corroborate visual comparisons. Our research attempts to develop such a system for tool mark evidence, proposing a methodology that objectively evaluates the association of striated tool marks with the tools that generated them. In our study, 58 primer shear marks on 9 mm cartridge cases, fired from four Glock model 19 pistols, were collected using high-resolution white light confocal microscopy. The resulting three-dimensional surface topographies were filtered to extract all "waviness surfaces"-the essential "line" information that firearm and tool mark examiners view under a microscope. Extracted waviness profiles were processed with principal component analysis (PCA) for dimension reduction. Support vector machines (SVM) were used to make the profile-gun associations, and conformal prediction theory (CPT) for establishing confidence levels. At the 95% confidence level, CPT coupled with PCA-SVM yielded an empirical error rate of 3.5%. Complementary, bootstrap-based computations for estimated error rates were 0%, indicating that the error rate for the algorithmic procedure is likely to remain low on larger data sets. Finally, suggestions are made for practical courtroom application of CPT for assigning levels of confidence to SVM identifications of tool marks recorded with confocal microscopy.

Statistical discrimination of footwear: a method for the comparison of accidentals on shoe outsoles inspired by facial recognition techniques.

In the field of forensic footwear examination, it is a widely held belief that patterns of accidental marks found on footwear and footwear impressions possess a high degree of "uniqueness." This belief, however, has not been thoroughly studied in a numerical way using controlled experiments. As a result, this form of valuable physical evidence has been the subject of admissibility challenges. In this study, we apply statistical techniques used in facial pattern recognition, to a minimal set of information gleaned from accidental patterns. That is, in order to maximize the amount of potential similarity between patterns, we only use the coordinate locations of accidental marks (on the top portion of a footwear impression) to characterize the entire pattern. This allows us to numerically gauge how similar two patterns are to one another in a worst-case scenario, i.e., in the absence of a tremendous amount of information normally available to the footwear examiner such as accidental mark size and shape. The patterns were recorded from the top portion of the shoe soles (i.e., not the heel) of five shoe pairs. All shoes were the same make and model and all were worn by the same person for a period of 30 days. We found that in 20-30 dimensional principal component (PC) space (99.5% variance retained), patterns from the same shoe, even at different points in time, tended to cluster closer to each other than patterns from different shoes. Correct shoe identification rates using maximum likelihood linear classification analysis and the hold-one-out procedure ranged from 81% to 100%. Although low in variance, three-dimensional PC plots were made and generally corroborated the findings in the much higher dimensional PC-space. This study is intended to be a starting point for future research to build statistical models on the formation and evolution of accidental patterns.