| 英文摘要 |
This study is divided into four experiments, using a Matrice 300 RTK UAV equipped with a Riegl D2M camera, conducted within the National Ilan University campus. A total of 10 control points and 20 checkpoints were established, and 3D modeling was carried out using iTwin Capture Modeler. The research focused on evaluating the accuracy and differences in 3D modeling results under the same flight altitude and forward overlap conditions, with side overlap rates of 40%, 60%, and 80%. In addition, comparisons were made between 3D models generated from vertical photogrammetry and oblique photogrammetry to assess the impact of different imaging approaches on modeling results. Furthermore, the study examined the effects of control point distribution and ground sampling distance (GSD) on the accuracy and quality of the generated models. The findings aim to determine the actual accuracy of oblique photogrammetry in 3D urban modeling and to propose optimized parameter settings for improving the practicality and reliability of models, providing valuable insights for future urban planning and management practices. Experimental Results: A higher side overlap rate leads to higher model accuracy, with 80% side overlap producing the best results. Oblique imagery yielded better modeling accuracy than vertical imagery, offering superior reconstruction of building exterior details. However, for applications with stricter requirements on orthophoto quality, single-lens vertical photography may be a more suitable option. Additionally, setting control points appropriately at higher elevations, using a smaller GSD, and lowering flight altitude can improve model accuracy. |