一种城区LiDAR点云数据的抽稀算法

当城区LiDAR点云数据密度较大时，存在大量的数据冗余，造成了计算量大、效率低、显示不便等一系列问题，使得建筑物的三维可视化及三维重建等应用受到较大挑战。针对该问题，结合泊松碟采样在测地空间中的地形自适应特点，提出了适用于城区LiDAR点云数据的抽稀算法。泊松碟采样随机将与已有采样点的测地距离大于某一阈值的点加入采样点集，并不断重复这一过程直至没有新的采样点加入为止。在此基础上，依据LiDAR点云数据的特点，定义了一种新的与所选点与其邻域内其他点间高度差标准差相关的加权测地距离，改进了泊松碟采样算法。该方法能有效调整城区建筑物的采样率，从而尽可能地保持建筑物的原始特征，并保留良好的可视化效果。四组对比实验结果表明了该算法的适用性及高效性。

A Thinning Algorithm of LiDAR Point Cloud Data in Urban Area

When the density of LiDAR point cloud data in urban area is high, there is so much data redundancy that a series of problems such as large computation, low efficiency, inconvenient display and so on arise, making the application of 3D visualization and 3D reconstruction of buildings more challenging. To solve this problem, a thinning algorithm suitable for LiDAR point cloud data in urban area is proposed, which combines the terrain adaptive features of Poisson disk sampling in geodesic space. Poisson disk sampling randomly add points whose geodesic distance is larger than a certain threshold to the sampling point set, and repeat this process until there are no new sampling points can be added anymore. On this basis, according to the characteristics of LiDAR point cloud data, a new weighted geodesic distance related to the height standard deviation of the points around the selected point is defined to improve the Poisson disk sampling algorithm. This method can effectively adjust the sampling rate of urban buildings, so as to keep the original features of buildings as much as possible, and keep good visualization effect at the same time. The experimental results of four sets of data demonstrate the applicability and efficiency of the algorithm.