Description of original award (Fiscal Year 2013, $150,000)
As submitted by the applicant:
The National Research Council has stated the need to establish a scientific basis demonstrating the validity of forensic science methods. An important task of such focus is the matching of specimens that at one time were of common origin. This task often uses subjective methods to arrive at a definitive conclusion. We propose to quantify the comparison of multiple torn and cut duct tape ends to provide an objective basis for determining a match. This research will address the issues of human contextual bias by using a quantitative algorithm, imaging software, and statistical methods to construct quantified profiles for torn tape specimens. Once we have these profiles we will evaluate the distribution of similarities among duct tape specimens. We will begin with an algorithm that uses a sum of the square differences to measure similarity. However, we will continue to evaluate other algorithms during this research.
We propose to use 2,400 torn and cut duct tape specimens that have previously been organized for end match research. This set of 2,400 tape specimens came from a successful prior NIJ sponsored and published research grants. The set comprises duct tape specimens torn by hand, cut by box cutters, cut by scissors, and torn by an Elmendorf tear tester. In subsequent research, we have developed a prototype system for quantifying matching of tape specimens. We will build on this research to create a database of digital images of all 2,400 tape specimens, to address such issues relating to tape specimen processing, to evaluate the distribution of the similarity measure, and to evaluate the use of objective criterion for matching in comparison to subjective analysis methods.
Using the prototype version of our algorithm we have successfully compared and uniquely identified a data base of 100 duct tape specimens (200 torn edges). This analysis requires the user to image the tape by using a scanner, align the image, select boundaries, and perform preprocessing to remove certain artifacts that interfere with the end match comparisons. This software was developed using MATLAB. However, our final version will be a standalone compiled program made available to analysts without the need for a license. The end product of this research will enable the forensic scientist to process a new torn duct tape specimen, obtain a quantitative measure of similarity, and compare its similarity to those in a data base. Furthermore, this concept of quantitative matching could logically be expanded to any areas wherein there is a physical match issue.