激光雷达技术和摄影测量方法在无人机数字测图中的精度对比分析

Comparison of UAV Digital Mapping Accuracy by Lidar Technology and Photogrammetry

为了对比分析雷达点云和摄影测量数字高程模型精度差异,以山西某地的数据为例,设计对比分析方案,通过数据计算的结果分析得到相关结论。具体方法为对两份点云数据分别做点云分类,分为植被、建筑物、地面点三类,注意两份地面点需要转换为参数相同的规则格网DEM。对比方法使用python编程实现,需要编程实现的步骤有数据整合、去除粗差、运用相关公式编写对比方法、输出结果。研究表明:Lidar点云和影像点云中不同地物的高程差异比较大,例如当不做点云分类时得到的均方根误差(E_(RMS))达到724 mm,而做过点云分类之后,植被层的E_(RMS)可以达到1138 mm,涨幅57%,建筑物层经计算降幅达85%,DEM降幅13%。可见在硬化表面地区,传统摄影测量得到的影像点云可以代替Lidar点云,其他复杂环境做精度对比分析时一定要先做点云分类,对比结果相对更合理。文章使用了一套成熟的点云精度对比方案,经过验证有很高的可行性,具有一定参考意义。

Taking the data of a particular place in Shanxi province as an example,we designed a comparative analysis plan and drew conclusions to compare the accuracy of digital elevation models by radar point cloud and photogrammetry from the data processing.The specific method is to classify the two copies of point cloud data into three categories:vegetation,buildings,and ground points(N.B.The two ground points’data need to be converted to a regular grid DEM with the same parameters).The comparison analysis is implemented by python programming,including data integration,removal of gross errors,comparison algorithm by relevant formulas,and output.The result shows that the elevation difference of objects in the lidar point cloud and the image point cloud is relatively significant.For example,without the point cloud classification,the root mean square error(E_(RMS))reaches 724 mm;with the point cloud classification,the E_(RMS) of the vegetation layer increases by 57%,which reaches 1,138 mm,the building layer decreases by 85%,and the DEM decreases by 13%.Therefore,in the hardened surface area,the image point cloud from traditional photogrammetry can replace the lidar point cloud;when comparing the accuracy of other complex environments,if the point cloud classification is performed first,the comparison results are relatively more reasonable.In addition,the article uses a mature accuracy comparison plan of point cloud,verified to be highly feasible,and has specific reference significance.