Abstract：The amount of vehicle-borne LiDAR point cloud data is tremendously and desultorily.Currently, there is still no one who complete and rigorous algorithm for point cloud auto-segmentation and model reconstruction of buildings, especially with the texture information of model reconstruction.In this paper, the author takes the building with a complex structure as an example.Firstly, considering the actual characteristics of the buildings, we take the automatic and manual methods to treat the process of segmentation and filtering by the principle of the consistency of concave and convex from simple to difficult.Secondly, we utilize the way of combining the point cloud data and CAD base map to make the quasi-three-dimensional reconstruction.Finally, the texture of models should be mapped and rendered.After all these work done, real three-dimensional model construction can be developed.The method has practical value to processing and modeling vehicle laser scan data.

Keyword：Vehicle-borne LiDAR;data segmentation;texture mapping;3D reconstruction;