Gaze-Controlled Simplification Of Triangle Meshes

In the field of computer graphics, there are kinds of representation of a 3D model, while the triangle mesh is one of the most popular ways because of its uncomplicated relational structure, abundant ability of expression, direct way of rendering and convenient scanning and acquisition. Further more, triangle mesh has a wide range of application in the field of virtual reality, animation,3D games, CAD and so on. Research on simplification of triangle mesh are of great sense with the consideration of easy conversion from other representations. With the development of hardware and modeling, mesh,simplification becomes more and more important in 3D applications.Gaze-controlled simplification which based its computation on observer's view can get better visual effect and higher efficiency than conventional method which based on geometric information. So, viewpoint based multiresolution rendering with appropriate error control can be a current method. Simultaneously, due to the dynamic viewpoint, the rendering should keep the pace with the movement. At the same time, we must smooth the discontinuous according to the viewpoint during real time rendering.Additionally, perceptually guided simplification is a hot area, which based on the achievements in visual sciences. When the detail on a 3D model is not perceptual, we simplify it till can be seen, or keep the detail. Perceptually guided simplification is more subjective compared to the conventional simplification methods which rely on the model's geometric topology information, for the participation of the observer's vision which guided the whole simplification process.The main research content and contributions of this paper can be summarized to the following two phases:1) Studying systematically and realized the dynamic gaze-controlled multiresolution rendering of a triangle mesh,summarizing existent work. With a preprocess to build a hierarchical mesh representation.Our method use quadric error control instead of simple edge collapsing to simplify the mesh during the preprocess in order to improve the visual effect, and insert the available information about the position and adjacency of the vertexes into the hierarchical structure. Then we search the hierarchical structure to render the appropriate vertexes according to the viewpoint. We also propose a advanced method to accelerate the mesh's continuity during the rendering in order to adjust to the movement of the viewpoint.2) We also advanced another simplification method which based on visual Perception. According to the contrast sensitivity function or CSF which quoted from the field of visual sciences, we analyze the complexity of the detail on the surface of a mesh model, compute its spatial frequency and contrast. Then we can judge whether the detail is perceptual, and make the decision to simplify it or not, and iterate the step until all the details of the model are perceptual. During the Perceptually guided simplification, our method add the viewpoint concerned region to the algorithm in order to improve the visual effect and make our method more flexible. Based on this work, we can further study how to improve it by accounting for textured models and dynamic lighting effect.