A Novel Approach to Gunshot Residue Pattern Interpretation Utilizing Digital Image-Based Analysis

Date of Submission


Document Type


Degree Name

Master of Science in Forensic Science


Forensic Science


Peter Massey

Committee Member

Timothy M. Palmbach

Committee Member

Martin D. Slade


Forensic ballistics, Forensic sciences, Criminal investigation

Call No. at the Univ. of New Haven Library

AS 36 .N29 For. Sci. 2014 no.1


Traditional methods of Gun Shot Residue (GSR) pattern interpretation primarily rely on methods such as trial and error, shot re-creation, and sometimes chemical composition analysis in order to estimate muzzle to target distances. In these investigations, the firearm in question must be test fired at different distances in order to create patterns to compare to the evidence. These methods rely on judgment and estimation without statistical or numerical support. In most cases, the actual firearm used in the crime is not even recovered to use for test firing. The goal of this study is to create and test an alternative method that may not only provide statistical support to a firearm/GSR pattern match, but may also aid in identifying or at least narrowing down possibilities for the gun that was fired, should it not be recovered.

In this study, a new technique for analyzing GSR patterns was developed and then preliminarily tested for its possible application to firearm distance and caliber determinations. In the first stage of sample collection, four different caliber firearms were used to create GSR patterns under controlled conditions, fired at distances ranging from six inches to sixty inches, onto white cloth targets. These patterns were then digitally photographed and uploaded to the software program ImageJ, a program developed by the National Institutes of Health that is used in photo analysis for various scientific applications. The photograph of each GSR pattern was analyzed in ImageJ by calculating the average grey value of different areas of the pattern, starting with a one centimeter square area directly surrounding the bullet hole (deemed “center” of the pattern) and progressing outwards in concentric squares to the edge of the target containing the pattern (up to 15 centimeters squared). These grey values were compared to the grey scale that was created in each picture using white and black reference squares photographed next to each pattern. The reference squares used were consistent across all photographs, but the grey reference scale was recalculated for each individual photograph in order to take into account lighting differences, a problem that would be very likely to occur in actual crime scene work. The result of these comparisons produced a “relative greyness” output value that could be translated into a mathematical measure of how much GSR was present in each measured region versus the amount of white space left untouched.

Several results were obtained from these initial studies. One such result is that there was a certain distance for each caliber at which there was not enough GSR residue that reached the target for it to produce any data using this analysis method. However, at the closer distances, preliminary analysis of the targets showed a clear trend of variation between the relative greyness gradient of GSR patterns at different distances for the same firearm. Although there was not enough data at this early stage to produce definitive statistics, using the limited samples there appeared to be a variance of anywhere from one to three standard deviations in some of the sequential distances tested with the same firearm. Additionally, each pattern was treated with either basic hairspray, Rustoleum fixative, or no fixative and there did not appear to be any significant difference in the stability of the pattern between the three methods.

In the second sample collection procedure, the same shooting conditions were utilized, but only a .45 caliber Smith and Wesson and a .40 caliber Glock were fired. Additionally the muzzle to target distances that were collected from were 6”, 12”, 18” and 24”, with 7 repeat trials for each caliber at each distance. The samples were photographed and uploaded by the same procedure used in part one, however the analysis varied slightly. Instead of square analysis regions the areas measured using ImageJ were circular, and instead of 1 cm intervals expanding out to up to 15 cm, the analysis was performed at 0.5 cm intervals starting from the center and continuing out to 7 cm in diameter. Also, since the preliminary studies showed no noticeable fixative effects, no fixative was used on any of the samples collected in part two. Once the data was analyzed from part two, a linear model was applied and statistical analysis performed, revealing a very high level of statistical significance.

This work has shown that this novel method of analysis has potential for application in real case scenarios involving GSR patterns at close ranges. Information from further studies on this method may allow for this digital image based process to address a large variety of other conditions that may affect GSR patterns that are created by firearms, and could potentially lead to the ability to create a database or simulation that can aid in pattern interpretation in everyday cases.