Rearch On The Modeling And Analysis Of The Tree-dimensional Point Cloud Data Of Trees

With the spread of digital forestry construction and improvement, the construction of forestry informationization has been advanced. In recent years,the research of building a more real 3d forestry visual model has become a hot spot in the research of 3d modeling research of forestry. As a new surveying and mapping technology, three-dimensional Laser Scanning has not only broadened the means of realizing the virtual reality and building of 3d forestry model but also has collected the 3d testing data of forest in a real, high-precision and digital way. It brings convenience for the survey of forestry resource, studying standard structure and 3d reconstruction of a single tree. Based on the 3d reconstruction of a single tree and point cloud data collecting from 3d laser scanning, the thesis realizes the noise reduction of original data and thus builds the Delaunay network model of a tree and triangulation network will output in the form of.X file. Therefore, a complete 3d forestry model has been formed.Taking advantage of point cloud data of tree as the data source, the thesis first preprocesses the 3d point cloud data. After comparing the de-noising algorithm of all kinds of discrete cloud points, the thesis used a more practical de-noising algorithm-bilateral filtering de-noising. By adjusting bilateral filtering space weight factor σc and gray scale weight factor σs, the thesis arrives at the optimal parameter combination, thus realizes the automatic de-noising of point clouds. The de-noising algorithm discussed in this paper is superior to the conventional algorithm and has the feature of high-efficient, convenient and no need of manual work.This paper is concentrated on the survey of different construction algorithms which used in the Delaunay Triangular Network and combined with the characteristics of multidimensional and complexity in tree point cloud data, then choose the most efficient Delaunay with incremental inserting improved algorithm to construct the model of Triangular Network. To make the constructional tree model more real and very efficient, this paper proposed the thought of re-order all discrete points, then optimizing the point location algorithm and LOP algorithm in aspects of data structures and efficiency of algorithms.This paper also focuses on the study of the format of.X file which used in the output of the tree modeling. As the.X file is a class of three-dimensional file format driven by template, the usage of.X file is convenient to the transformation of the model data, and process of the Render Back-Ends.One of the most efficient measure on constructing complex 3D model, especially the 3D modeling of defoliation tree, is the model reconstruction of the cloud data with the three-dimensional scanning technique. The survey of the algorithms in this paper provide a new method of the high-efficient modeling of complex tree model. These algorithms will possess strong theory and application with great values.