Noise reduction and modeling methods of TLS point cloud based on R-tree

3D terrestrial laser scanning (TLS) appears in 1980s. As a new high-tech, TLS can fast obtain the 3D coordinates of objects on the ground surface through high-speed laser scanning. However, it is still an urgent problem that how to deal with the high density and unorganized point clouds, including reducing noise and modeling these mass data. The main purpose of this paper is to research how to reduce noise for the huge point cloud data, and establish TIN efficiently for modeling accurately after the data preprocess. This paper utilizes R-tree which has been widely used in building index to build the dynamic index structure, and then reduces noise points through K-Nearest Neighbor Query (KNN) based on R-tree. Finally it adopts triangulation growth method based on KNN to establish TIN, realizes the modeling process for point cloud data, and proves the valid and feasibility of the algorithm by experiments.