With previous experience in automated bullet identification
research using acquisition of topography images and the cross
correlation score as a similarity metric [11], we try to address the
aforementioned questions about statistics and subjectivity, and to
provide a method for automation of CMS as the similarity metric. A
computerized identification method based on 3D topography
measurement and CMS criteria is described in this report. Since
both the bullet features and the topography acquisition procedure
are three dimensional, the bullet data enable a model to be built to
fit the 3D CMS criteria. CMS values are automatically calculated
from surface topography images for a set of unknown bullets
compared to a set of knowns. The CMS results are compared with
the true results revealed subsequently. The paper is organized as
follows: a brief introduction of the CMS counting model is
presented in Section 2 including image preprocessing, feature
profile extraction, and striae-matching determination. The experimental
results and statistical analysis are given in Section 3. In
Section 4 we conclude with an evaluation of all the results.