Rapid and Physically-Based Gaussian Splatting of Unknown Space Objects in Low Earth Orbit

Tae Ha Park

Keywords: Gaussian Splatting, 3D Reconstruction, Rendezvous, Proximity Operations

Abstract: This paper presents a physically-based 2D Gaussian Splatting (2DGS) pipeline for novel view synthesis and 3D reconstruction of unknown spaceborne targets in Low Earth Orbit (LEO). In this setting, accurate reconstruction requires not only geometric fidelity but also photometric consistency, as appearance cues play a key role in estimating camera poses from image observations. However, existing approaches often rely on simplified lighting assumptions that limit their ability to model real space environments. To address this, the proposed method explicitly incorporates illumination priors by modeling the Sun as a directional light source and the Earth as a secondary, predominantly diffuse illumination term, capturing the dominant lighting effects in LEO. A deferred shading formulation, combined with an analytical Bidirectional Reflectance Distribution Function (BRDF) leveraging the inherent surface normals in 2DGS, enables efficient material estimation and rapid reconstruction under varying illumination conditions. Experiments on a high-fidelity synthetic dataset of the Hubble Space Telescope, featuring realistic and challenging materials, illumination, and sensor effects, demonstrate strong material consistency and rendering quality, particularly for challenging appearance effects such as Earth-induced reflections on specular surfaces.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Paper Acceptance: No

Submission Number: 31