顾及点密度与未知角分辨率的地面点云分类

为了解决地面激光点云角分辨未知和点密度变化的问题, 提出了一种顾及密度变化与未知角分辨率的地面激光点云分类方法。采用随机邻域分析的角分辨率估算法改进传统点密度计算方法, 结合角分辨率提出顾及密度变化的格网特征提取方法, 并将本文中所提出的方法在3组数据上进行了实验验证。结果表明, 该角分辨率估算方法的估算误差小于0.002°, 能够准确地估算点云角分辨率; 与传统点密度特征相比, 本文中提出的相对投影密度可以提高点云分类的整体精度, 以及在车、杆状物等地物类别上的分类效果。该方法能准确估算点云角分辨率, 以较高精度实现点云分类, 可为大规模地面激光点云的密度自适应处理提供参考。

Classification of terrestrial point cloud considering point density and unknown angular resolution

In order to solve the problems of unknown angular resolution and point density variation of terrestrial laser point cloud, a classification method considering density change and unknown angular resolution was proposed in this paper. To improve the traditional point density calculation method, the angular resolution estimation method of random neighborhood analysis was presented. Then we combine angular resolution to propose a grid feature extraction method which takes density variation into account. The proposed method was tested on three datasets. The result shows that the error of our method is smaller than 0.002°, which can accurately estimate the angular resolution. And compared with traditional density feature, our method can improve the overall accuracy of point cloud classification, and perform well in the extraction of cars and pole. The angle resolution can be accurately estimated with this method, and the point cloud can be classified with higher accuracy, which can provide a reference for density adaptive processing of large-scale terrestrial laser point clouds.