基于格网划分的海量地形数据三维可视化

3D Visualization of Massive Terrain Data Based on Grid Partition

首先按照格网划分方式对研究区域进行分割 ,采用与视点相关的静态LOD模型 ,在格网块内以四叉树进行管理 ,根据误差以二元三角树方式进行LOD模型预处理 ,并采用三角形条带进行若干不同层次的LOD模型文件组织 ;然后按照格网块进行视景体投影裁剪 ,结合动态内存调用、多线程管理的数据引擎方法以及视觉光滑处理 ,实现了海量地形数据的三维可视化及漫游 实验结果表明 ,文中算法具有处理数据量无限制、效率高。

A 3D visualization and roaming algorithm of massive terrain data based on grid partition, dynamic RAM and view dependent static LOD model is presented. Area of interest is first subdivided into hierarchical quadtree of square zones. According to the predefined approximation errors of LOD, each quadtree is preprocessed into binary triangle tree and stored as LOD model data files in the form of triangle stripes. View frustum projection culling, data engine of dynamic RAM with multi-thread management and geomorphing are applied to realize 3D visualization and roaming of terrain. Experimental results show that the data processing capacity is unlimited with high efficiency and good effect.