Development of a Camera Placement System for Construction Site Monitoring

Abstract

Vision-based monitoring techniques, not necessitating additional sensors or tags to be installed except cameras, have been widely introduced to support construction site monitoring. However, challenging issues still remain as to how to determine the locations of cameras to improve monitoring performance under the limitations in installation and visibility. A suitable camera placement must be determined before installation to obtain high monitoring performances. To help making reasonable decisions, this research proposes a camera placement system that includes specific installation criteria and quantitative analysis metrics. To establish the camera placement system, 3D modeling of the site was first carried out to build a virtual environment based on the actual images collected by Unmanned Aerial Vehicles (UAVs). Second, factors to consider derived from site investigations and related researches were reflected in the 3D models of the site as installation criteria. Third, virtual cameras with intrinsic parameters such as focal length, field of view, and visible distance were placed. Lastly, quantitative assessment was made by calculating the overall coverage of the multiple cameras. Introducing a systematic and quantitative system for camera placement by considering site conditions, this study contributes to improving construction site monitoring performances by successfully obtaining quality image data for vision-based monitoring.