MoonGS: High-quality Representation of the Lunar Surface via Gaussian Splatting Using Robust Depth Features from Image Pairs

2026-10-05

AI Summary

This study addresses the challenge of 3D reconstruction from sparse lunar images characterized by weak textures and low overlap. We propose the first feed-forward 3D Gaussian Splatting framework tailored for lunar scenes. By predicting Gaussian primitives and rendering novel views through a single forward pass from only two input views, our method eliminates the need for per-scene optimization. Furthermore, it integrates depth features from vision foundation models with semantic priors and introduces an entropy-guided heuristic resampling strategy to effectively enhance geometric consistency under sparse observations. Experimental results demonstrate that the proposed approach achieves state-of-the-art performance on both the LuSNAR dataset and Chang'e mission data, yielding a 4.9 dB improvement in PSNR, a 40% reduction in LPIPS, and sub-second inference speed.

Abstract

High-quality 3D reconstruction of lunar terrain from sparse rover images is indispensable for autonomous lunar exploration, but remains challenging because viewpoint overlap is insufficient, surface textures are weak, and data volume is limited. We propose MoonGS, the first feed-forward 3D Gaussian Splatting framework tailored to lunar scenes. Given only two input images, MoonGS predicts pixel-aligned Gaussian primitives in a single forward pass and renders photorealistic novel views without any per-scene optimization. MoonGS (i) adopts an adaptable backbone design that seamlessly integrates advanced vision foundation models to extract robust depth features; (ii) integrates semantic priors in two manners: merging semantic cues with visual features to refine Gaussian parameter estimation, and adopting a semantic ranking loss that regularizes background depth; and (iii) employs an entropy-guided heuristic resampling strategy to augment sparse observations by selecting the most informative distant viewpoints with negligible overhead. Experiments on the LuSNAR benchmark and our synthetic weak-texture MoonBlender dataset show that MoonGS surpasses state-of-the-art feed-forward NeRF/3DGS baselines by +4.9 dB PSNR, +0.29 SSIM, and 40\% lower LPIPS while maintaining sub-second inference. Furthermore, we validate the broad applicability of our framework by demonstrating that it effectively leverages state-of-the-art backbones, including VGGT, to significantly boost performance. Qualitative evaluations on Chang'e mission imagery also show the best visual quality among compared methods, indicating robustness on real lunar data. The source code and dataset are publicly available at https://github.com/InRobots/MoonBlender.

Problem

• 3D reconstruction
• lunar terrain
• sparse images
• weak texture
• Gaussian Splatting

Innovation

• 3D Gaussian Splatting
• Feed-forward Reconstruction
• Depth Features
• Semantic Priors
• Lunar Surface