2D-SuGaR: Surface-Aware Gaussian Splatting for Geometrically Accurate Mesh Reconstruction

2026-05-01

AI Summary

This work addresses the sensitivity of existing 2D Gaussian splatting methods to Structure-from-Motion (SfM) initialization, which often leads to geometric distortions in complex scenes. To mitigate this issue, the authors propose a depth-guided Gaussian initialization strategy that integrates monocular depth and surface normal priors to enhance the quality of the initial geometry. Furthermore, they introduce a clustering-based mechanism for the automatic removal of degenerate Gaussians, significantly improving reconstruction robustness. Evaluated on the DTU dataset, the proposed method substantially outperforms current approaches, achieving state-of-the-art mesh reconstruction accuracy while preserving high-fidelity novel view synthesis.

Abstract

3D Gaussian Splatting (3DGS) has emerged as a powerful technique for generating photorealistic renderings of a scene in real-time. However, the volumetric nature of 3DGS limits its ability to accurately capture surface geometry. To address this, 2D Gaussian Splatting (2DGS) was proposed to enable view-consistent and geometrically accurate surface reconstruction from multi-view images. However, 2DGS can be sensitive to the initialization of the Gaussian primitives. Reliance on Structure-from-Motion (SfM) initializations, which can produce poor estimates on challenging image sets, may lead to subpar results. In this work, we enhance 2DGS by incorporating monocular depth and normal priors to improve both geometric accuracy and robustness. We propose a depth-guided initialization strategy for Gaussians and introduce a clustering-based technique for pruning degenerate Gaussians. We evaluate our method on the DTU dataset, where it achieves state-of-the-art results in mesh reconstruction while preserving high-quality novel view synthesis.

Problem

• 2D Gaussian Splatting
• surface reconstruction
• initialization sensitivity
• geometric accuracy
• Structure-from-Motion

Innovation

• 2D Gaussian Splatting
• depth prior
• surface reconstruction
• geometric accuracy
• initialization strategy