Reconstruction, Based On Neural Network Classifier To Distinguish The Nurbs Surface Terrain

Terrain surface model has its important and wide--ranged application background in virtual reality. Terrain surface reconstruction has wide application prospect in the geographic information systems, computer-aided design and graphics, computer modeling, reverse engineering, virtual simulation and other applications. Terrain surface reconstruction also has been a hot issue.At present, it is very difficult to build a corresponding three-dimensional model from a terrain image with mass of characteristics, by manually identify the slope shape of images to create the model based on the current status of the computer's development. So in general the increase speed method from artificial intelligence aspects, by computer algorithms to identify the intelligent terrain analysis and ultimately create a surface topography are an important method to this problem.In this dissertation, focused on the research of the surface model rebuilding in the virtual realistic environment, combine the picture processing and classification technique, through a high degree of map and the generation of NURBS surface, make the fluctuations re-surface on the screen. The Terrain reconstruction system established based on the theory of BP neural network and Reconstructing B-spline Surface. The results make sure that this method can meet the construction of virtual scenes, so do its validity and practicability.