Building Damage Visualization through Three-Dimensional Reconstruction Using Images and Point Clouds

Abstract. Assessing building damage caused by natural disasters, such as earthquakes and typhoons, involves an initial survey to measure wall inclination and determine risk levels. Traditional plumb line surveying methods can be influenced by external factors, such as wind and vibration, and require physical contact with the damaged buildings, posing a risk of collapse during the survey. This study proposes a method using a camera capable of simultaneously capturing three-dimensional (3D) point clouds and images. This method extracts the entire wall surface of a building through point cloud processing and calculates its inclination. In the experiments, the accuracy of the system was first verified using a simple storage shed, followed by validation of its effectiveness on an actual house. The results demonstrated that, compared with the conventional plumb line method, the proposed method could remotely measure the building without physical contact, ensuring the safety of surveyors. The proposed method achieved measurements with less error and provided a measurement accuracy comparable to or better than the plumb line method. The assessed inclination was used to determine the risk level of the wall surfaces, enabling an intuitive and visual evaluation by color-coding the entire wall surface according to the risk level.