Were these bullets fired from the same gun? Are these markings on the cartridge casings close enough to be a match? Forensic examiners are faced with questions like these, and in this presentation, we discuss some of the shortcomings in the current methods, illustrate how new technology can improve the comparison process, and describe a statistical approach to quantifying the similarity between cartridge casings as well as the uncertainty in making comparisons.
We quantify the distance between images using a modified version of the Kullback-Leibler divergence that is symmetric and uses local averaging. Using actual images of cartridge casings from numerous guns, we implement our distance statistic to study its variability when comparing cartridge casings fired by different guns and the same gun. We also address the variability as a function of which region of the image is selected.
|