一种高效的大规模三维地形遥感影像纹理映射方法

An Efficient Mapping Method For Large-scale 3D Terrain Texture of Remote Sensing Images

针对大规模虚拟战场地理信息建模与绘制的需求,研究提出一种可充分发挥GPU和多CPU核微处理器处理能力的、高效的大规模三维地形遥感影像纹理映射方法。该方法直接利用GPU进行高效的纹理压缩解压,采用基于多CPU核微处理器的多线程并行调度策略,实现了视点相关的纹理数据的高效预取和调度,并根据等地理长度映射原理实现了单一精度的大规模遥感影像纹理数据在大规模LOD地形上的精确匹配。实验表明,此方法可有效提高大规模三维地理信息场景绘制的速度,纹理和地形的匹配精度高,增强了三维地理信息漫游的真实感。

Aiming at the demand of Modeling and Rendering a large-scale virtual battlefield Geographic Information,a highly efficient large-scale 3D Terrain texture mapping method of remote sensing image that gives full play to GPU and multi-core CPU microprocessor processing capacity is presented in the paper.By directly using GPU for texture compression and decompression and using Multithreading parallel scheduling strategy fur view-dependent texture data prefetching and scheduling based on multi-core CPU microprocessor,Also,the exact match of single accuracy large-scale remote sensing image data is implemented on large-scale texture LOD terrain,according to the principle of equal geographical Length mapping.Experimental results show that the performance of rendering in the large-scale 3D terrain information scene can be effectively improved and the reality of 3D geographic information can be further enhanced by making the texture and the topography exactly match.