Assessment of the Positional Accuracy of Digital Elevation Model Derived from Low-Cost UAV การประเมินความถูกต้องเชิงตำแหน่งของแบบจำลองระดับสูงเชิงเลขที่ผลิตจากอากาศยานไร้คนขับต้นทุนต่ำ

Abstract

The objective of this research was to evaluate the positional accuracy of a digital elevation model derived from low-cost Unmanned Aerial Vehicle in the Recreation Park of Burapha University, Mueang district, Chonburi province using 6 ground control points and 24 checkpoints obtained through Real-Time Kinematic (RTK) surveys. The results indicated that the Root Mean Square. Error (RMSE) value at the confidence level of 95%, without using GCPs had a horizontal positional accuracy of 0.073-0.246 meters and a vertical positional accuracy of 0.270-0.429 meters. These values meet the standards set by the National Standard for Spatial Data Accuracy (NSSDA) and the American Society for Photogrammetry and Remote Sensing (ASPRS). Based on these results, the researchers recommend creating a photographic map at a scale of 1:4,000. Additionally, contour range data should be generated with a minimum interval of 100 centimeters. When GCPs were utilized in the process, the RMSE value at the confidence level of 95%, demonstrated the horizontal positional accuracy ranged from 0.054-0.093 meters, while the vertical positional accuracy ranged from 0.114-0.333 meters. These values also complied with NSSDA and ASPRS requirements, allowing for the creation of a 1:1,000 scale photographic map. Additionally, contour range data can be generated with a minimum interval of at least 50 cm. According to this study, digital elevation models from low-cost UAVs can be used both without and with GCPs for applications that do not require very detailed and precise verticality values, such as building elevation surveys, tree height, or the height of large billboards, etc., because the difference in elevation values between features within the study area is still within acceptable criteria.