A Generalization Method Of 3D Buildings Based On WebGL

With the rapid development of open 3D data and Web 3D graphics technology,3D visualization of buildings based on WebGL is applied more and more widely in many fields.Howerver,the graphics and computing capability of browsers still lags behind the rapid growth of 3D data,it is difficult to rendering these 3D models in high speed;Besides,a map with lots of 3D building models has bad visual effects that has seriously hindered people's cognition of spatial information.So the generalization of 3D building models is essential,but most of the existing generalization algorithms are defined based on geometric features of 3D bulidings,not considering visual effects and spatial cognition of people.Generalizing 3D buildings data which will be visualized based on WebGL in a way accord with human visual perception and spatial cognition,is worth studying,which can reduce the amount of redundant information of 3D data,ensure good visual effect of map,improve the rendering speed and accelerate cognitive efficiency of spatial information.This paper firstly discusses the relationship between map generalization and spatial cognition,points out that only verified by spatial cognition theory,can we ensure the consistency of spatial information and maintain the visual characteristics of the buildings.Secondly,this paper analyses the factors affecting huaman visual perception and spatial congnition in the process of 3D buildings visualization and calculates these factors in methods of geometric and spatial anglysis models;Then,we generalize and simpilify 3D buliding models based on spatial clustering and pattern recognition of 3D city bulidings.Above of all this paper put forward a cognition-based generalization algorithm of 3D city bulidngs,the main contents include:1.Defining and quantifying factors affecting human visual perception in the process of visualizing 3D biuldings,including 3D bulidings' topological features,geometric features,distance,size,shape,orientation,semantic features and visual effect changes due to diffrent viewports in 3D scene;And then calculating proximity relationship between buildings based on these quantitative parameters.2.Creating a minimum spanning tree(MST)from the proximity graph based on visual perception distance,and after that doing generalization and simplification of 3D building models on the basis of spatial clustering and distribution pattern results.Finally,on the basis of the current WebGL map visualization framework,this paper designs a WebGL 3D building real-time generalization framework which is suitable for online mapping applications.Experiments using open 3D data form OSM prove the effectiveness of the proposed method.