CSAFE researchers have recently published a new paper in the Journal of Forensic Science addressing the need for objectivity and repeatability in firearm analysis techniques. Our team has created a fully automated, open‐source method for comparing breechface marks on cartridge cases.
In current practice, when examiners retrieve cartridge cases from a crime scene they can enter it into a national database, running a search on a subset of images matching some criteria, such as caliber of gun and type of crime (e.g. homicide). This search returns the top ranked potential matches.
Xiao Hui Tai, a CSAFE graduate student working under the direction of CSAFE Co-Director Dr. William F. Eddy said, “Examiners would look at these images on the screen, and if they think there is a possibility of a match they can retrieve the physical cartridge cases and examine them under a comparison microscope. They could then potentially testify in court about their findings.”
“Examiners might say things like ‘to a reasonable degree of scientific certainty’, but in reality they are basing their conclusions on their own training and experience,” said Tai. “We’re trying to come up with objective way of doing comparisons, not necessarily to replace these opinions, but to attach a numerical measure of similarity which might corroborate these conclusions, and back up such statements.”
This CSAFE method focuses on 2-dimensional optical images and consists of an image analysis algorithm to compute a measure of similarity. “Our method improves on existing methodology by modifying pre-processing steps,” said Tai.
Instead of selecting breechface marks manually, potentially introducing examiner-level variability in the comparisons, CSAFE researchers can now automatically select the breechface marks. This leads to a reduction in the subjectivity of the process.
“Our new method also removes the effects of differences in depth of the cartridge case base that are circular in nature; this can also be described as circular symmetry. The CSAFE algorithm now corrects the resulting differences in brightness that previously affected the quality of the image,” said Tai. These new steps result in a demonstrable improvement in accuracy on images obtained from controlled test fires.
Tai describes how the CSAFE method is increasing transparency in the field of firearms and toolmark evidence analysis. “A lot of methods available are not open source,” she said. “A big advantage of methods being open source is the ability for all parties to see where they succeed and where they fail.”
CSAFE advancements in breechface analysis enable fast and reliable matching of cartridge images, decreasing the level of effort by examiners. The developed statistical model can also be used for blind verification in crime laboratories, and for reporting a weight of evidence. “Our idea is for this automated method to allow examiners to verify their decisions using an algorithm. In the last step of our analysis we also compute the probability of obtaining the result just by chance, which is something that can be reported in courts,” Tai said.
Tai and the CSAFE team are working on scaling up the method to perform searches on a large database. “A lot of groups use methods that are not fully automatic and focus on a small number of accurate comparisons. We are currently working on extending this method so that it can perform faster and not lose too much accuracy. Another thing we are working on is applying methods to 3-D topographical data in addition to 2-D images,” Tai said.
As a publicly available resource, the forensic science community is able to utilize the CSAFE breechface analysis method to continue improving objectivity and accuracy in this field. The CSAFE algorithm can be found in the Datasets and Tools section of the CSAFE website.
Learn more about CSAFE research on firearms and toolmarks on our website, and review our news article about new bullet matching technology. Discover how open-source methods benefit forensic evidence analysis on the CSAFE blog.
