| 摘要 |
Currently, the aerial survey system of low-altitude unmanned aerial vehicles (UAVs) has been widely used in acquiring digital map 4D products, mapping, digital linear maps, and other aspects. However, there are problems, such as low precision and weak practicability in constructing digital elevation model (DEM) products through the data collected using consumption level UAVs. Therefore, improving the accuracy of DEM products obtained by consumption level UAVs is a crucial and complex issue in the research of UAV aerial survey systems. In precision elevation measurement, the geodetic height of a certain number of ground points with reasonable distribution in the region is often obtained first. Then, the normal height of the ground points is obtained by leveling, and the elevation residual value surface of the region is fitted. Finally, the normal height of the points to be solved in the region is obtained by fitting the elevation residual surface. Therefore, the elevation residual fitting method was used to improve the accuracy of consumer UAV DEM products in this study. First, a high-quality ground point cloud was obtained by constructing the gradient filtering-cloth simulation filtering (GF-CSF) model. Second, an abnormal elevation fitting residual DEM model was constructed. Lastly, the final DEM was obtained using the DEM difference method. The experimental results show that among the 20 random sampling inspection points, the average elevation residual was 2.3 mm, and the root mean square error (RMSE) was 16.7 mm after the DEM accuracy was improved by the method. The average elevation residual without improving the DEM accuracy was 28.6 mm, and RMSE was 33.7 mm. |
