Photogrammetric Point-Cloud Replicability When Documenting Forensic Archaeological Scenes under Variable Lighting Conditions

Abstract

Forensic archaeological scenes involving human skeletal remains in wooded environments can be challenging to documentutilizing close-range photogrammetry (CRP) due to the complex nature of outdoor scenes. Previous research has demonstrated that changinglighting conditions can negatively affect three-dimensional (3D) model quality. The purpose of this research was therefore to test theimpact of variable lighting on the replicability of 3D point clouds using CRP in a wooded environment. One scattered scene was createdusing a composite human skeleton and several clothing items. The scene was photographed three times during one day to capture changinglighting conditions: at 9:45 am (Model 1), at noon (Model 2), and at 2 pm (Model 3). Photographs were collected freehand from multipleview angles using a Sony α7 III camera with a fixed wide-anglelens, and the models were processed using Agisoft Metashape Professional.All three models achieved a total scale bar error of less than 1 mm and therefore met the accepted standards for crime scene mappingbest practices. The dense point clouds were then analyzed using CloudCompare to assess point-cloud replicability between modelpairs. The Multiscale Model to Model Cloud Comparison (M3C2) tool was used to calculate signed distances between point-cloud pairs.Also, histograms that display these point-to-pointdeviations were generated for each comparison, and a Gaussian distribution was fittedto each histogram. Deviations between point clouds were minimal, indicating that CRP-generated point clouds are replicable under changinglighting conditions, as well as other environmental variables, such as mild wind conditions and complex ground surfaces. Thus, it isrecommended that forensic archaeologists incorporate CRP into their documentation protocol.