MonoGSDF: Exploring Monocular Geometric Cues for Gaussian Splatting-Guided Implicit Surface Reconstruction

MCML Authors

Kunyi Li

→ Group Federico Tombari
 Computer Aided Medical Procedures & Augmented Reality

Zeyu Chen

→ Group Enkelejda Kasneci
 Human-Centered Technologies for Learning

Nassir Navab

Prof. Dr.

Core PI

Computer Aided Medical Procedures & Augmented Reality

Abstract

Accurate meshing from monocular images remains a key challenge in 3D vision. While state-of-the-art 3D Gaussian Splatting (3DGS) methods excel at synthesizing photorealistic novel views through rasterization-based rendering, their reliance on sparse, explicit primitives severely limits their ability to recover watertight and topologically consistent 3D this http URL introduce MonoGSDF, a novel method that couples Gaussian-based primitives with a neural Signed Distance Field (SDF) for high-quality reconstruction. During training, the SDF guides Gaussians' spatial distribution, while at inference, Gaussians serve as priors to reconstruct surfaces, eliminating the need for memory-intensive Marching Cubes. To handle arbitrary-scale scenes, we propose a scaling strategy for robust generalization. A multi-resolution training scheme further refines details and monocular geometric cues from off-the-shelf estimators enhance reconstruction quality. Experiments on real-world datasets show MonoGSDF outperforms prior methods while maintaining efficiency.

misc LNC+25

---

Preprint

Nov. 2025

---

Authors

K. Li • M. Niemeyer • Z. Chen • N. Navab • F. Tombari

---

Links

arXiv

---

In Collaboration

 Google

---

Research Areas

 B3 | Multimodal Perception

 C1 | Medicine