Based On The Point Of The Shape And Rendering Technology

With the rapid developing of Computer Graphics, 3D computer images have penetrated many application. With continuously developing of computer modeling and rendering technology of requirement, the computer models the size of modeling scene becoming bigger, the details of scenery become more abundant. Especially, with large-scale popularization and application of 3D digital scanner, the number of grid points which are got by scanning is reaching at billion, the details and form which are got by scanning are becoming more abundant, the requirement of efficient and treating with the big model of the new method is increasing. If using traditional trangle-based rendering method, current hardare cannot reach the requirement of in-time rendering of such complex models. Recently, a novel technique point-based rendering become more and more hot. With the increasing research on point-based rendering, point-based rendering will become the research hot point of computer graphics.This paper first introduces the whole flows of point-based obtaining, disposing and rendering, the properties of point-base denoting surface, and the present popular several point-based modeling methods.(1) Basing on Kohonen neural network's self-organizing feature map of being capable of keeping topologic structure, surface is reconstructed from scattered data points. A rectangle mesh reconstruction approach based on the self-organizing feature map neural network is developed. The inherent topologic relations between the scattered points on the surface are learned by the self-organizing feature map neural network. The weight vectors of the neurons on the output layer of the neural network are used to approximate the scattered data points. By this approach, it is not only to approximate the scattered data points and the surface which is reconstructed by this method can be as base surface for further process, but also the dense scattered data points can be reduced to the reasonable scale. And constructing topology shape which keeps original data collections, this reduces the number of original data collections and enlarges application field of neural network surface reconstruction.(2) Basing on RBFNN' s the approaching any no-linear function by arbitrary precision, powerful antinoise and the capability of repair and so on, adopting RBFNN model to reconstruct free-form surface, we constitute RBFNN model fitting to surface...