Development of CG Technologies Based on the Geometric Model Represented by a Set of Points

点击翻译按钮获取中文摘要

In this study, we developed several kinds of CG techniques related to a set of points as a fundamental geometric model: modeling and rendering techniques, mutual transformation techniques among a set of points and another model, especially a polygon model. The following models are regarded as a set of points : 2D scalar grid fields such as depth images obtained by a range scanner and DEM(Digital Elevation Model), 3D scalar grid field such as medical images, voxel spaces, 3D textures and point clouds mainly obtained by a particle simulation. The approaches of rendering techniques are classified into those: to apply a conventional polygon rendering to a polygon model transformed from a set of point, to apply point-based rendering directly to a set of points, and to apply point-based rendering to a set of points transformed from another type of model. Especially, the latter approach is expected as a most possible method having the abilities of LOD(Level Of Detail) and anti-aliasing indispensable to render broad natural landscape sceneries including huge amount of complex objects like trees.In this study, we developed the following techniques for handling a set of points1.Modeling techniques for depth images(1.1)Position matching algorithms for integrating multiple depth images(1.2)Algorithms for making the distribution of points uniform(1.3)An algorithm for adapting geometric primitive to depth image using Genetic Algorithm2.Rendering algorithm for voxel spaces(2.1)An edge preserving algorithm for smoothing normal3.Surface generation algorithms from a set of points(3.1)A level-set algorithm using velocity fields(3.2)A recursive algorithm for finding a polygon hull4.Point-based rendering(4.1)Point-based rendering for simulating first order scattering(4.2)An algorithm for transforming a polygon model to a 3D texture(4.3)An algorithm for transforming a polygon model to a point cloud