【作者】 张慧杰； 孙吉贵； 孙红光； 罗娜；

【Author】 ZHANG Hui-jie~(1,2) SUN Ji-gui~1 SUN Hong-guang~2 LUO Na~2 1 (College of Computer Science and Technology,Jilin University,Changchun 130012) 2 (College of Computer Science,Northeast Normal University,Changchun,130024)

【机构】 吉林大学计算机科学与技术学院； 东北师范大学计算机学院；

【摘要】 本文提出了一种基于聚类分析的外存地形模型实时可视化算法,该算法根据静态误差建立外存 LOD 模型, 目的是将不参与绘制的高程数据在载入内存前剔除掉。本文采用不完全四叉树以多分辨率的形式存储该模型,同时根据静态误差和空间位置关系将四叉树中的节点进行聚类划分。每个聚类存储四叉树节点的索引信息,并且类内节点采用 Hilbert 空间填充曲线的形式进行排序,在具体绘制时形成较长的三角形条带,从而进一步提高了索引和实时绘制的效率。本文对 GTOP30数据进行了实验,结果取得了良好的绘制和漫游效果。更多还原

【Abstract】 This article presents a real-time terrain visualization algorithm based on clustering analysis.The algorithm established out-of-core LOD models on the basis of static error,which eliminate the elevation data that will not participate in rendering before the data are loaded into memory,and stored it with incomplete quad-tree in the form of multi-resolution model.Based on the static error and the space location relations will the nodes in quad-tree be divided into clusters,and each cluster stored index information of those nodes.Because those nodes are sorted in the form of Hilbert space filling curve,a longer triangle strip can be formed so as to further enhance the efficiency of indexing and real-time rendering.In this article, the experiments on GTOP30 DEM data were performed,and the good results about rendering and roaming have been achieved.更多还原