计算机应用 ›› 2010, Vol. 30 ›› Issue (12): 3292-3294.

• 虚拟现实与模式识别 • 上一篇    下一篇

地形可视化中的改进Geoclipmap算法

张建廷1,刘福太2,艾祖亮1   

  1. 1. 海军航空工程学院
    2.
  • 收稿日期:2010-06-22 修回日期:2010-08-02 发布日期:2010-12-22 出版日期:2010-12-01
  • 通讯作者: 张建廷

Improved geoclipmap algorithm for terrain visualization

  • Received:2010-06-22 Revised:2010-08-02 Online:2010-12-22 Published:2010-12-01
  • Contact: Zhang Jian-Ting

摘要: 在大地形实时绘制中,大规模的地形数据和有限的硬件数据通信带宽是限制地形绘制效率的主要原因。在Geoclipmap算法的基础上,通过使用几何场景图(GSG)组织结构提高数据外存加载效率,在mipmap棱锥生成过程中采取sinc滤波方法进行重采样,避免地形细节丢失。为减少CPU到图形处理器(GPU)的数据流量,提出一种基于层次包围球的二级视锥体裁剪技术,并将法线的生成放到GPU的片段着色器中。实验结果表明,算法保持地形真实感,并有效提高绘制效率,能满足大地形的实时渲染要求。

关键词: 地形可视化, 细节层次, 剪切图, 重采样, 视锥裁剪, 包围体

Abstract: In the real-time rendering of large terrain, it is the contradiction between large terrain data and the limited data communication bandwidth of the hardware that affects the rendering efficiency. The survey was based on the Geoclipmap algorithm and some improvements were made to solve the problem. During the data preprocessing, a Geometry Scene Graph (GSG) structure was used to organize the terrain data in order to improve data loading speed, and the implementation used a sinc function as the filter in the generation of mipmap levels to avoid detail of terrain surface being flattened. After that, it is the rendering process. A two-level view frustum culling was presented which was based on the bounding spheres hierarchy and normal computation was put in the fragment shader of Graphic Processing Unit (GPU), which reduced the data stream between CPU and GPU. Just as the simulation results show, the algorithm can keep the fidelity of terrain and be effective to meet the needs of large terrain rendering.

Key words: terrain visualization, Level of Detail (LOD), clipmap, resampling, view frustum culling, bounding box