High-precision 3D Modeling of Construction Waste Pile Bodies by Integrating Multi-source Data

Abstract. The instability of construction waste pile bodies, as an increasingly concerning disaster, poses significant risks to the safety of people's lives and property. Currently, there is limited research on high-precision 3D modeling techniques for construction waste pile bodies, which significantly hinders the accuracy and reliability of early detection and risk assessment of pile body instability. Therefore, constructing high-precision 3D models of construction waste pile bodies using multi-source data plays a crucial role in improving the accuracy and timeliness of early warning systems for pile body instability, offering significant theoretical research and practical application value. Building 3D models of construction waste pile bodies solely based on unmanned aerial vehicle (UAV) oblique photography data faces challenges, such as various degrees of data voids and insufficient model completeness, and few methods currently address the construction of 3D models of construction waste pile bodies through the fusion of multi-source data. This paper attempts to use the Iterative Closest Point (ICP) algorithm, combining SLAM laser point cloud data with UAV oblique photography data to fill data voids, and utilizes the fused point cloud to reconstruct the triangulation network, achieving the transformation from 3D point cloud models to 3D surface models. On the basis of texture mapping, a high-precision 3D model of the construction waste pile body is constructed. The research results show that the 3D model of the construction waste pile body, integrated with multi-source data, has a planar error of 0.0187m and an elevation error of 0.0368m, meeting the corresponding model accuracy requirements. It can more realistically restore the fine 3D features of the construction waste pile body, effectively compensating for the shortcomings of single-source data in 3D modeling of construction waste pile bodies, providing a new method for the 3D model reconstruction of construction waste pile bodies, and offering effective data support for construction waste research.