Atmosphere-aware photoclinometry for pixel-wise 3D topographic mapping of Mars

High-resolution topographic mapping is essential for scientific investigations and operational exploration of planets, such as Mars. Photoclinometry, which uses light scattered from a surface to reconstruct 3D topography, can retrieve subtle topographic details from monocular images. However, its performance is affected by the atmosphere of Mars, which alters surface reflectance mechanisms due to the scattering and absorption by aerosols. Therefore, specific treatments accounting for these atmospheric effects are needed to enable photo-clinometric mapping of Mars. This paper presents a novel photoclinometric approach incorporating a radiative -transfer model of atmospheric scattering effects for pixel-wise 3D reconstruction of the Martian surface. The approach requires a high-resolution image, a corresponding coarse-resolution digital elevation model (DEM), and information of the optical depth as inputs. A radiative transfer model adapted to the Martian atmosphere is used to account for the atmospheric effects. The approach also allows directly adopting optical depth estimates from a global database (e.g., the Mars Climate Database). The approach was validated using different types of Mars orbiter images collected by cameras on-board the Mars Reconnaissance Orbiter and the Tianwen-1 Orbiter. The results indicate that the approach can achieve a geometric accuracy (in terms of root-mean-squared error (RMSE) of the elevation) of approximately 2 pixels of the image resolution and significantly enhances topographic de-tails. In addition, we evaluated the approach using different settings for the optical depths and spatial resolutions of the input DEMs. The results show that overestimating the optical depth leads to overestimation of topographic amplitudes (e.g., deeper craters). On the other hand, underestimating the optical depth by the same amount as overestimation leads to a smaller increase in the RMSE. Moreover, coarsening the resolution of an input DEM increases the RMSE of the photoclinometric results. Nevertheless, photoclinometry improves both the RMSE and the resolution of the input DEM. This new approach serves as an effective means for applying photoclinometry for pixel-wise 3D topographic mapping of the Martian surface. This will facilitate exploitation of the large number of high-resolution monocular images of Mars in 3D topographic mapping of the planet.