Three-dimensional model processing, plant modeling and scene realistic redering are important in the field of computer graphics research. They can find their applications in virtual scenes, 3D movies, 3D games, digital city, garden planning, forestry researchment, and so on. In the past few decades, many experts and scholars in the three-dimensional model processing and modeling and rendering plants have published a great deal of fruits, but there are very few works for topiary modeling . Topiary plant is an important part of landscape. Topiary modeling is helpful for mesh analysis and constraint modeling. Therefore, both from a practical point of view or perspective of theory, topiary plant modeling research has important significance. In addition, with that 3D laser scanner continuously improves its performance, more and more 3D point cloud data are acquired. So it is necessary to manage these new data and to reconstruct objects from them. Study on the three-dimensional geometric model processing in this thesis is as a theoretical basis for modeling plant. Plant modeling and scene modeling are as the direct application of the theoretical analysis and processing of three-dimensional geometric model. The visualization of the plant model and plant scene is done by achieving realistic rendering. The main contents of the thesis are as follows: analysis and processing on the discrete point set, point cloud-based reconstruction of the plant, forest scan data segmentation, tree modeling with crown shape constraints, multi-resolution modeling, realistic rendering and scene rendering. Main contributions of this paper are: 1) A method called Chord And Normal vectors (CAN) has been proposed for estimating the principal curvatures and the principal directions. The main thaulenge is from that without analysis formula and topology information. The principal curvature can not be calculated with classical geometry method. We use position and normal vector of neighbor point to extimate the normal curvature, and then obtain the principal curvature with an optimum method. The main feature of this method is robust. This new algorithm is used to extract local differential properties, including saddle regions, sharp ridge regions and branch growth direction. Those results are the basic of plant reconstruction and forest segmentation. 2) We present a tree reconstruction approach based on scan point cloud. It is difficult to reconstruct small branches for all published meth...
修改评论