运用矩阵结构的可并行地形层次细节算法
A Parallel LOD Algorithm Using Matrix Structure
-
摘要: 针对ROAM算法因分块间的耦合关系不适合大规模地形渲染, 以及算法递归实现导致绘制效率低下的问题, 结合嵌套网格算法压缩比高的优点, 提出一种运用矩阵结构的可并行计算的地形层次细节算法.采用矩阵结构消除地形分块间的耦合关系, 以解决大规模地形的分块渲染问题;用对矩阵的遍历代替递归操作来提高算法的执行速度;最后将计算并行化处理, 并利用CUDA进行加速.实验结果表明, 文中算法适用于大规模地形的绘制, 比传统ROAM算法的速度提高了2~20倍.Abstract: The ROAM algorithm is not suitable for out-of-core rendering because of inter-block coupling, and is inefficient for recursion.With a high compression ratio of nested regular grids algorithm, the paper proposes a parallel LOD algorithm using matrix structure.This algorithm adopts matrix structure to eliminate the inter-block coupling, and is suitable for out-of-core rendering.It also eliminates the recursion to improve performance, proposes the parallel algorithm and uses CUDA to accelerate the processing.Experimental results show that the algorithm is suitable for visualization of out-of-core terrain data, and has a 2~20 times speedup compared with traditional ROAM algorithm.