Abstract: With the improvement of data acquisition ability of 3D laser point cloud, building model reconstruction and elevation mapping based on 3D laser point cloud has become a common practice in engineering applications. 3D laser point cloud data can reflect abundant and intuitive details of buildings. However, the processing of massive 3D point cloud data brings great challenges to building models. The contour points of buildings are extracted by cross-cutting the 3D laser point cloud data of buildings. The TSP algorithm base on genetic algorithm are used to process the contour points and obtain the equation coefficients of the facades of buildings. Finally, the construction of building models and the detailed vector data of the facades of buildings can be obtained. The experimental results show that this method can better realize the construction of LOD1 building model, providing a basis for building model construction of higher (LOD3) level.

Key words: 3D laser point cloud, crosscutting, building model, genetic algorithm, TSP algorithm