Multi-sensor points cloud data fusion for 3D building models: A case study in Halong City, Vietnam

Abstract

Three-dimensional building models play a crucial role in urban planning, emergency response, disaster management, and decision making. Multi sensor data fusion has recently attracted significant interest in the Geomatics research community. This approach addresses the limitations of individual sensors, allowing for the creation of comprehensive 3D models of structures and improving object classification. This study focuses on developing approaches that combine various geospatial technologies to produce a complete 3D model of common urban architectures, including high building, neighboring villa, and individual home. This research used flexibly employ UAV aerial imagery, ground photography, and terrestrial LiDAR scanning to collect the necessary information for constructing complete 3D models of each characteristic urban structure. Different point cloud data sets will be processed, merged, and used to generate the complete 3D models. The experimental results have produced complete 3D models with accuracy achieved with Δx; mΔy; mΔz all below 10 cm for the experimental buildings. With the accuracy of the 3D models, it is entirely possible to achieve accuracy in horizontal plane and height for the 1:500 scale terrain map.