Visualization of quality parameters for classification of spectra in shooting crimes

Abstract

X-ray emission spectra from gunshot residues (GSR) are evaluated in order to assign them to a certain ammunition brand. Classification is based on regularized discriminant analysis (RDA) providing dass information as well as additional parameters for judging the quality of an assignment. For visualization of measures such as the posterior probability, distance to centre or inverse Mahalanobis distances different approaches are explored and their common interpretation is evaluated. Copyright © 2008 John Wiley & Sons, Ltd.