基于Mesh模型的LOD2.2级城市三维模型智能化构建OA

To address the problems that oblique photography mesh models are difficult to directly extract building footprints and suffer from severe data redundancy,this paper proposes an intelligent method for constructing LOD2.2-level 3D city models.Based on a region-growing algorithm,through adaptive seed point diffusion,intelligent boundary determination,and automatic topology repair,the automated and precise segmentation of building footprints is achieved;from three geometric dimensions—feature points,contour lines,and triangular meshes—systematic lightweight processing of individual building models is carried out using techniques such as curvature-driven decimation,contour regularization,and coplanar mesh simplification.Results show that this method enables the automatic transformation from'real-scene 3D meshes'to'structured individual models',generating streamlined LOD2.2 models that integrate complex roof forms with regular geometric features,effectively addressing the challenges of low-efficiency semantic segmentation and large model data volumes,and providing a reference for intelligent 3D city modeling.