一种基于机载点云的DEM生成方法研究

Study on a DEM Generation Method Based on Airborne Point Cloud

摘要:

DEM即数字高程模型，是通过有限的地形高程数据实现对地面地形的数字化模拟。DEM具有许多生产方式，作为直接获取对象表面点三维坐标的现代对地观测技术，机载激光扫描在DEM构建方面具有很大优势。因此，本文研究了一种基于机载激光点云数据的DEM生成方法，该方法的关键在于机载点云的地面滤波处理。本文提出了一种改进的渐进三角网滤波方法，通过计算各点坡度以及邻域范围内的高差最大值，进行直方图统计分析，实现高度及角度阈值的自适应估计。将本文结果与人工滤波及布料滤波方法进行对比分析。实验结果表明，本文方法的结果更加贴近人工滤波处理效果，可有效提高DEM生成的精度。

Abstract:

In this paper, a method based on 3D laser point cloud data has been introduced, which eliminates the influence of ground object points on data interpolation through progressive triangulation filtering, and interpolates the filtered point cloud to obtain DEM based on the inverse distance weighting principle. The key of this method is the filtering effect of the progressive triangulation algorithm. In this paper, a histogram is generated based on the normal vector of each point and the maximum height difference of each point within the neighborhood. The method of preserving most point clouds is adopted for parameter selection and filtering. This result has been compared and analyzed with manual filtering and fabric filtering. The experimental results show that the asymptotic triangular network filtering effect after parameter estimation is closer to the effect of manual filtering. By comparing the DEM generated by artificial filtering and progressive triangulation filtering, the RMSE value is calculated, and the results show a small difference, and the rationality of this method has been proved.