Engineering and Applied Science Research (Jul 2022)
Toward 3D reconstruction of damaged vehicles for investigating traffic accidents in Thailand using a photogrammetric approach
Abstract
Traffic accident investigation by the police forensic science department in Thailand is essential to determine their causes. The collection of forensic evidence from damaged vehicles after an accident plays a significant role in damage assessment and collision trajectory analysis. This study employs a photogrammetric approach using an off-the-shelf mobile device and free software for the three-dimensional (3D) reconstruction of damaged vehicles to investigate traffic accidents. In this study, the iPhone XS Max mobile phone camera was used for image acquisition of close-range photogrammetry and videogrammetry. After image capture, 3D models of forensic evidence were automatically reconstructed from the imagery using the COLMAP open-source software that provides user-friendliness to non-experts. The 3D models of the deformed vehicles were later used to analyze the damage and collision trajectory of a traffic accident. The results showed that for accuracy assessment of the 3D model test car, the values of root mean square error (RMSE) obtained from still images, video, and video with a stabilizer were 2.5, 3.1, and 2.4 cm, respectively. The completeness of the generated 3D model obtained from still images provided greater clarity than videos with and without stabilizers, respectively. Therefore, the photogrammetric approach using a mobile device played a significant role in the 3D reconstruction of the forensic evidence used for traffic accident investigations, providing essential 3D information for court trial reports.
