Airborne And Terrestrial Lidar Data Fusion For Building 3D Reconstruction

Building information has a great influence on the sustainable development of society and the construction and planning of the city. How to collect this information quickly and effectively has become the focus of current research. This paper focus on how to collect each plane of building and how to reconstruct the three-dimensional modeling of buildings. With multi-source LiDAR cloud point data fusion and multi platforms and angles, the plane filtering algorithm of global threshold can extract the building plane targets more accurately. At last, this paper complete three-dimensional white mode of building construction. The main results in this paper are as follows:(1) Discussed the method of semi-automatic point clouds registration fusion which used of the feature line. Focus on the semi-automatic registration and fusion method, and improved it. Combined with the special nature of the experimental data, took the flatness method using vector constraint, extraction linear features based on geographic rough positioning, through the method of search feature line and semi-automatic data registration and fusion. Through the experiments, compare the different characteristic of registration and fusion effect. On the comparative analysis of before and after improved filtering algorithm based on the plane fitting slope. Using the method of geographic rough positioning and feature line location can be better to realize the fusion and registration of ground and airborne LiDAR point cloud data, the effect of registration and fusion is closely related with the characteristic line selection.(2) Improved plane fitting using slope filtering algorithm. In the summary of researches by scholars at home and abroad based on related issues, the threshold processing part of the algorithm is improved according to the algorithm. Focus on the problem of threshold is fuzzy, pre classification of the building plane, using total least squares method of global threshold estimation to improve the accuracy of plane fitting. Compare the filtering plane fitting surface extraction and the global threshold of normal constraint; Summed up the advantages and disadvantages of these methods; Analysis the accuracy and the feasibility of the algorithm. Using total least squares method of the global threshold of plane fitting filtering algorithm can optimize the plane fitting, and can improve the accuracy of plane fitting to a certain extent.(3) Realized 3D reconstruction of experimental data. The experiment use Alpha Shape algorithm to extraction top surface information of the building from airborne LiDAR point cloud data. Use vector classification constraints extraction window information. Useg patch semantic detection of building facade information from the ground LiDAR data.3D reconstruction of complete geometric calibration is achieved after building. Spatial geometry calibrated building plane has characteristics of higher accuracy, stronger adaptability, better reflects the detailed 3D information building. Combination of ground and airborne LiDAR point cloud data reconstruction of buildings can be more fine reflects the building 3D information, but its degree of automation is still need to improve.Research shows that:semi-automatic point clouds registration method of feature line can integrate ground and airborne cloud data better, but this method is influenced by the selection of feature line; total least squares plane fitting filtering method can improve the accuracy of the improved plane fitting; extracting contour information of combined ground and airborne cloud point can reflect the 3D information of building.