Accuracy of ray pinning compared to model-based image matching for forensic investigations

Forensic investigations often require accurate placement of objects or persons in an incident scene in order to establish the most likely scenario of how events transpired. This can be accomplished through ray pinning, a technique in which control points on a model of an object of interest and a 2D image of the incident scene are correlated to optimally match the location of the object in 3D space to its location in the 2D image. Alternatively, a technique referred to as model-based image matching (MBIM) relies on the acuity of an operator's vision to manually manipulate the location of the model until the operator judges that the model of the object is overlaid as accurately as possible on the 2D image of the object, as represented in the 3D space. The purpose of this study is to compare the accuracy of ray pinning to MBIM in positioning an object using 2D images from video frames. A simulated scene, in which a Blueguns rifle had been placed on the ground, was captured in videos taken by three stationary cameras placed in different locations. The position and orientation errors for the rifle placement was calculated for ray pinning and MBIM. Both techniques employed a 3D scan of the scene used to calibrate the cameras. The results of statistical analysis showed that MBIM was significantly more accurate in positioning the rifle than ray pinning, although the two techniques were equally accurate in orienting the rifle.