抄録
The recent progress of mid/ long-range laser scanners enables to capture large-scale point-clouds in short time. However, point-clouds typically contain many large noises and outliers, therefore, reliable smoothing methods are very important. Although the moving least-squares or the moving robust estimate is powerful smoothing tools, they are very time-consuming for large-scale point-clouds. In this paper, we propose a GPU-based method to accelerate the moving robust estimate. We implemented it on a GPU and evaluated its performance. The result shows that our GPU-based method is much faster than CPU-based smoothing, and it can be applied to large-scale point-clouds.
