A 3D modelling method for road based on vehicle laser point cloud

To address the challenges pertaining to the sluggish 3D reconstruction of the road surface, excessive workload, and limited adaptability, this study undertook the feature extraction of road point cloud data exhibiting diverse linear shapes, leveraging vehicle-mounted laser point cloud data. Additionally, a novel approach was devised to automatically construct the road triangulation network, thereby achieving a high-precision digital restoration of the road surface. Initially, the data attributes of the laser point cloud obtained from road vehicles are considered, and a data preprocessing technique is devised using the Alpha Shapes algorithm and the point cloud grid thinning algorithm. This method aims to retain crucial information pertaining to the elevation of the road surface, linear boundaries of the road surface, and skeletal features of the road surface. Subsequently, a technique is devised for constructing a Delaunay triangulation network using ordered point cloud data, based on the boundary points and feature points of the road surface. Additionally, an investigation is conducted on a method for constructing a road surface triangulation network with road boundary constraints, enabling the identification and removal of triangles located outside the road surface, thereby achieving precise restoration of the road surface. Ultimately, the assessment of the generated 3D pavement model's quality validates the method's feasibility and accuracy.