The increasing use of UAV-based LiDAR systems for high-resolution mapping highlights the need for reliable,field-validated accuracy assessment methods.This study presents a practical technique for evaluating geometric...The increasing use of UAV-based LiDAR systems for high-resolution mapping highlights the need for reliable,field-validated accuracy assessment methods.This study presents a practical technique for evaluating geometric and radiometric performance using georeferenced,high-reflectivity foil targets.The method enables precise extraction of target centers and correction of systematic georeferencing errors through 3D transformation.The approach was applied at the Tora Cement Factory in Cairo,Egypt—an industrial site with complex topography—using a DJI Matrice 300 RTK UAV equipped with the Zenmuse L1 LiDAR sensor and Zenmuse P1 photogrammetric camera.Three test flights were performed at altitudes of 50 m(nadir and oblique)and 70 m(oblique),with a high-resolution Structure-from-Motion(SfM)point cloud generated for reference.After transformation,the global RMSE of the LiDAR dataset was reduced to approximately 2.8∼3.2 cm,improving upon the raw uncorrected accuracy of up to 4.6 cm.Surface-wise comparisons showed RMSEs of 3.1 cm on flat areas,3.8 cm on rugged terrain,and 4.5 cm on vertical structures.Additionally,the RGB data embedded in the LiDAR point cloud exhibited a systematic spatial offset between 18 and 43 cm,with an average internal standard deviation near 5 cm,indicating a potential limitation for radiometric applications.The proposed method offers a cost-effective,accurate,and repeatable solution for UAV LiDAR validation and supports operational deployment,quality assurance,and system calibration in real-world scenarios.展开更多
文摘The increasing use of UAV-based LiDAR systems for high-resolution mapping highlights the need for reliable,field-validated accuracy assessment methods.This study presents a practical technique for evaluating geometric and radiometric performance using georeferenced,high-reflectivity foil targets.The method enables precise extraction of target centers and correction of systematic georeferencing errors through 3D transformation.The approach was applied at the Tora Cement Factory in Cairo,Egypt—an industrial site with complex topography—using a DJI Matrice 300 RTK UAV equipped with the Zenmuse L1 LiDAR sensor and Zenmuse P1 photogrammetric camera.Three test flights were performed at altitudes of 50 m(nadir and oblique)and 70 m(oblique),with a high-resolution Structure-from-Motion(SfM)point cloud generated for reference.After transformation,the global RMSE of the LiDAR dataset was reduced to approximately 2.8∼3.2 cm,improving upon the raw uncorrected accuracy of up to 4.6 cm.Surface-wise comparisons showed RMSEs of 3.1 cm on flat areas,3.8 cm on rugged terrain,and 4.5 cm on vertical structures.Additionally,the RGB data embedded in the LiDAR point cloud exhibited a systematic spatial offset between 18 and 43 cm,with an average internal standard deviation near 5 cm,indicating a potential limitation for radiometric applications.The proposed method offers a cost-effective,accurate,and repeatable solution for UAV LiDAR validation and supports operational deployment,quality assurance,and system calibration in real-world scenarios.
