多时段不同定位精度的无人机影像点云的对比分析

利用两个不同地形特征的实验区(平地和山地),使用大疆精灵4 RTK设计实施多种无人机影像采集和处理方案(3个时段、使用或关闭集成的RTK功能、数据处理时是否引入地面控制点),两个区域共采集12个架次的无人机影像，生成24个不同类别的影像点云模型.比较、验证不同条件下无人机影像点云模型的精度，评价不同方案的精度差异.结果显示，无人机能在一定程度上仅凭借集成的RTK功能实现免像控采集高精度数据，但仍然建议同时使用RTK模块采集数据，并引入少量地面控制点进行数据处理，此时获取的数据精度最好；中午时段的影像数据质量最佳，成果精度最高，两个实验区中午时段垂直方向最小平均差可达0.01～0.02 m,其他时段结果次之；地表特征对影像点云模型精度也有影响，地势平坦、地表特征丰富有利于模型重建，提高模型质量.

Comparative analysis of UAV dense-matching point cloud from images captured with different positioning accuracy in multiple time periods

Using two experimental areas with different characteristics (flat and mountain), DJI Phantom 4 RTK was used to design and implement a variety of drone image acquisition and processing schemes (3 periods, use or close the integrated RTK function, whether the data processing Introducing ground control points), a total of 12 UAV images were collected in the two areas, and 24 image point cloud models of different categories were generated. Compare and verify the accuracy of the UAV image point cloud model under different conditions, and evaluate the accuracy differences of different schemes. The results show that the UAV can only rely on the integrated RTK function to collect high-precision data without ground control points to a certain extent, but it is still recommended to use the RTK module to collect data at the same time, and introduce a small number of ground control points for data processing, the data obtained at this time The accuracy is the best; the image data quality during the noon period is the best, and the result accuracy is the highest, followed by the results in other periods; the surface features also affect the accuracy of the image point cloud model. The flat terrain and rich surface features are beneficial to model reconstruction and improve model quality.