Analysis of Focal Length Variations by UAV Fligth methods and Oblique Imagery Effects Seungchan Lim. 1, Dongyu Kim.2, and Chuluong Choi.3*
1) Division of Earth and Environmental System Science (Major of Spatial Information Engineering), Pukyong National University, Republic of Korea
2) Division of Earth and Environmental System Science, Pukyong National University, Republic of Korea
3) Division of Earth and Environmental System Science (Major of Spatial Information Engineering), Pukyong National University, Republic of Korea
*cuchoi[at]pknu.ac.kr
Abstract
Focal length (FL) varies depending on the type of UAV (Unmanned Aerial Vehicle) used for aerial photography. The FL, defined as the distance from the lens center to the image sensor, includes not only the original focal length but also the calibrated focal length (CFL), which is utilized during the photogrammetry process. The CFL can vary depending on distortion correction and flight conditions. The Mavic 3 Enterprise (M3E) automatically captures oblique images in the final phase of Grid missions using the Pilot 2 app to perform altitude optimization. This study aims to analyze changes in FL under various flight methods and processing conditions, and to examine the impact of including oblique images on mapping accuracy. The experiment was conducted using the M3E at Pukyong National University (Yongdang Campus), South Korea, on June 23, 2025. A total of three flights were conducted using Grid, 8-directional, and Dome methods. The maximum flight altitude was set to 35 meters, with altitude variations applied in the 8-directional and Dome methods. Five flight methods were compared: Grid (with oblique images), NOgrid (without oblique images), 8-directional, Dome, and ALL. Each method was analyzed under four processing conditions: Original, GCP, Cut, and GCP+Cut. FL analysis was performed using Agisoft^s Metashape. The results showed that the NOgrid, which excluded oblique images, showed significant fluctuations and instability in FL values compared to the Grid. In particular, the F_error for Grid remained stable between 0.062 and 0.065 under all conditions, whereas NOgrid showed a sharp increase up to 3.019 before GCP correction, which was reduced to 0.501 after GCP application. These findings indicate that using GCPs is effective in reducing focal length correction errors, and their role becomes even more critical in the absence of oblique imagery.
Keywords: Focal Length (FL)- Unmanned Aerial Vehicle (UAV)- Calibrated Focal Length (CFL)- Ground Control Point (GCP)
