Digital terrain maps (DTMs) are essential for planetary exploration, supporting landing site selection, terrain-relative navigation, and geological analysis. A new arXiv paper introduces OrTeN, a neural inverse graphics method designed to reconstruct both terrain and the trajectory of a line-scan camera.
The paper frames OrTeN as an alternative to existing state-of-practice reconstruction techniques for creating DTMs from satellite imagery. By jointly estimating terrain and camera trajectory, the method aims to solve the reconstruction problem in a unified framework.
The abstract does not include quantitative results or implementation details, but the title and framing indicate a focus on neural inverse graphics as a way to combine geometric and photometric cues. The approach may be relevant to future planetary missions that rely on accurate terrain models.