研究目的
To investigate empirically how the accuracy of a 3D SfM model varies with the location and number of control points used for geo-referencing.
研究成果
The accuracy of UAV-SfM photogrammetry depends on the number and location of GCPs used in the bundle adjustment. Using a higher number of GCPs improves the accuracy, and an even distribution of GCPs is crucial. The study also found that accuracy is overestimated when measured only using control points rather than independent check points. The maximum accuracy achievable is related to the ground sample distance (GSD) of the project.
研究不足
The study was conducted using a specific UAV and camera setup over a particular type of terrain. The results may not be directly applicable to other UAV systems or different terrains. The study also did not explore the impact of different SfM algorithms or camera calibration methods on the accuracy of the model.
1：Experimental Design and Method Selection:
The study used a small fixed-wing UAV plane equipped with a Samsung NX500 camera to capture images over a highland and mining area. The images were processed using Agisoft PhotoScan Professional software.
2：Sample Selection and Data Sources:
The test site was a coal mining area with a wide variation in relief, covering 1225 ha. 102 ground control points (GCPs) were placed evenly throughout the area.
3：List of Experimental Equipment and Materials:
The UAV was equipped with a Samsung NX500 camera with a Samsung NX 20 mm f/
4：8 lens, and a Javad Triumph-1 survey-grade GNSS receiver was used to determine the coordinates of the GCPs. Experimental Procedures and Operational Workflow:
The area was divided into three parts, and two photo flights were carried out for each subzone. A total of 2535 pictures were taken. The images were processed using Agisoft PhotoScan Professional software, and the accuracy of the model was evaluated using both control points and independent check points.
5：Data Analysis Methods:
The accuracy of the model was evaluated by calculating the root mean square error (RMSE) for both control points and check points.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
