APPLICATION OF SEGMENTATION ALGORITHMS IN TERRESTRIAL LASER SCANNING TO ENHANCE THE ACCURACY OF AREA CALCULATIONS OF ARCHITECTURAL STRUCTURES

Authors

DOI:

https://doi.org/10.32782/forensic.science.2026.4.16

Keywords:

3D modeling, oriented bounding box, positioning accuracy, photogrammetry, traffic accident reconstruction

Abstract

This article addresses the problem of improving the spatial positioning accuracy of vehicles in video frames under challenging perspective angles. The aim of the research is to develop a methodology for the precise localization of vehicle dimensions, which neutralizes the «streamlining effect» of modern ergonomic car bodies and eliminates systematic errors that occur during the calculation of kinematic motion parameters. The methodological basis of the study comprises general scientific and specialized methods, specifically the principles of projective geometry, computer vision techniques, computational photogrammetry, and the analysis of expert practice. The research is based on the analysis of video recordings of road traffic accidents where vehicles are captured at an acute angle to the camera’s optical axis, making the direct visual identification of their extreme dimensional points impossible. The proposed method relies on a comprehensive combination of 3D modeling tools, the analytical determination of spatial shooting conditions, and the back-projection of a verified 3D model onto the video frame plane. A key element of this approach is constructing a rigid framework around the model in the form of a virtual, semi-transparent oriented bounding box (OBB), whose planes strictly align with the extreme structural boundaries of the vehicle’s body. This framework serves as a geometric reference that enables the precise restoration of vehicle parts that are objectively hidden due to the camera angle. The scientific novelty lies in the development of an end-to-end method for aligning parametric 3D models with monocular video footage based on calculated vanishing points. This shifts the object positioning process from subjective visual estimation to rigorous geometric verification. The conclusions emphasize that applying the proposed method minimizes the human factor and establishes strict technological boundaries for the investigation. The method provides an objective geometric basis and verified initial data, forming an essential foundation for any subsequent expert calculations in forensic investigations. The practical value of this work lies in creating a reliable applied toolkit for forensic experts, which significantly enhances the evidential weight, objectivity, and verifiability of expert conclusions during legal proceedings related to road traffic accidents.

References

1. Choi Y., Kong S.-H. Robust video-based vehicle speed estimation for occluded scenes for forensic analysis of traffic accidents. Multimedia Tools and Applications. 2026. Vol. 85. Art. 90. DOI: https://doi.org/10.1007/s11042-026-21222-9.

2. Kocur V., Ftáčnik M. Detection of 3D bounding boxes of vehicles using perspective transformation for accurate speed measurement. arXiv. 2020. Preprint. DOI: https://doi.org/10.48550/arXiv.2003.13137.

3. Sochor J., Juránek R., Herout A. Traffic surveillance camera calibration by 3D model bounding box alignment for accurate vehicle speed measurement. Computer Vision and Image Understanding. 2017. Vol. 161. P. 87–98. DOI: https://doi.org/10.1016/j.cviu.2017.05.015.

4. Ospina-Bohórquez A. et al. Comprehensive Forensic Tool for Crime Scene and Traffic Accident 3D Reconstruction. Algorithms. 2025. Vol. 18, no. 11. Art. 707. DOI: https://doi.org/10.3390/a18110707.

5. Dubská M., Herout A., Sochor J. Automatic camera calibration for traffic understanding. Proceedings of the British Machine Vision Conference (BMVC 2014). BMVA Press, 2014. DOI: https://doi.org/10.5244/C.28.42.

Published

2026-11-18

How to Cite

Pereidenko, A., & Moroz, Y. (2026). APPLICATION OF SEGMENTATION ALGORITHMS IN TERRESTRIAL LASER SCANNING TO ENHANCE THE ACCURACY OF AREA CALCULATIONS OF ARCHITECTURAL STRUCTURES. Forensic Science, (4), 125–132. https://doi.org/10.32782/forensic.science.2026.4.16

Issue

Section

Статті