Impact of atmospheric correction on 3D reconstruction using GF-7 satellite imagery

The GF-7 satellite possesses stereoscopic imaging capability, which can generate surface point clouds through dense matching in 3D reconstruction. Conventional photogrammetric processing directly uses uncalibrated digital number (DN) images, and the radiation measurement errors in the original images may affect the 3D reconstruction results. Atmospheric correction of remote sensing images eliminates atmospheric noise via radiative transfer algorithms, providing clear surface reflectance (SR) images. This paper conducts comparative 3D reconstruction experiments in urban areas using GF-7 stereo images before and after atmospheric correction. The results demonstrate that atmospheric correction is valuable for 3D reconstruction of GF-7 stereo images, improving building contour preservation, point cloud quantity, and density. Thus, it is recommended to incorporate atmospheric correction into the photogrammetric processing workflow.