Research On The Technology Of LOD In Virtual Reality

LOD technology, which is one of the most efficient methods to accelerate scene rendering, is the main topic of paper dissertation. In this dissertation, the generation and dynamic control of models at multiple levels of detail is studied.The Basic concepts, primary characters and key techniques of VR are described firstly. In the mean time some methods to accelerate scene rendering are introduced briefly. Then a systematic survey on LOD technology is introduced, which includes the concepts, the research content and main algorithms of LOD.The dissertation comes from two aspects as follows:Firstly, paper proposes and implements a new progressive mesh simplified method. In this method, edge collapse is also mesh simplification's basic operation, but the cost of edge collapse is measured by the product of the edge length and surface curvature and quadric error metric which is easy to be estimated and can reflect geometric character of the mesh model. The collapsed edge is chosen based on the principle of the least cost of the edge collapse, and the objective vertices generated by edge collapsing are confined to the subset of the original mesh vertices, so geometric information of new vertex needn't to be stored. The experimental results indicate that this algorithm is easy to implement, runs fast and occupy small storage. moreover it can guarantee a high resemblance of the geometric characters between the simplified model and the original model.Secondly, View-dependent dynamic LOD can represent different parts of the model with meshes at different level of detail. A new view-dependent dynamic LOD algorithm based on a binary tree has been presented in dissertation. Improving the structure based on the binary tree and reducing the depth of the tree structure.Design a query function based on view-frustum and surface orientation. Collapse or split the node of the binary tree based on the view_dependent criteria. Adjust the level of detail of different parts of model. Based on the differents of view parameter, the algorithm uses operation of the binary vertex tree to represent different parts of model with meshes at different level of detail. The experiment results indicate that this scheme may attain a much higher frame rate in scene rendering as well as maintain the visual quality.