3DGSR: Implicit Surface Reconstruction with 3D Gaussian Splatting

TL;DR

This paper introduces 3DGSR, a novel implicit surface reconstruction method combining 3D Gaussian Splatting with an SDF, enabling detailed, efficient, and high-quality 3D reconstructions with improved supervision and regularization techniques.

Contribution

The paper proposes integrating an implicit signed distance field with 3D Gaussian Splatting, along with a differentiable SDF-to-opacity function and volumetric rendering regularization, to enhance detail and efficiency in 3D surface reconstruction.

Findings

Achieves high-quality 3D surface reconstruction with detailed features.

Maintains the efficiency and rendering quality of 3D Gaussian Splatting.

Outperforms leading surface reconstruction methods in accuracy and speed.

Abstract

In this paper, we present an implicit surface reconstruction method with 3D Gaussian Splatting (3DGS), namely 3DGSR, that allows for accurate 3D reconstruction with intricate details while inheriting the high efficiency and rendering quality of 3DGS. The key insight is incorporating an implicit signed distance field (SDF) within 3D Gaussians to enable them to be aligned and jointly optimized. First, we introduce a differentiable SDF-to-opacity transformation function that converts SDF values into corresponding Gaussians' opacities. This function connects the SDF and 3D Gaussians, allowing for unified optimization and enforcing surface constraints on the 3D Gaussians. During learning, optimizing the 3D Gaussians provides supervisory signals for SDF learning, enabling the reconstruction of intricate details. However, this only provides sparse supervisory signals to the SDF at locations occupied by Gaussians, which is insufficient for learning a continuous SDF. Then, to address this limitation, we incorporate volumetric rendering and align the rendered geometric attributes (depth, normal) with those derived from 3D Gaussians. This consistency regularization introduces supervisory signals to locations not covered by discrete 3D Gaussians, effectively eliminating redundant surfaces outside the Gaussian sampling range. Our extensive experimental results demonstrate that our 3DGSR method enables high-quality 3D surface reconstruction while preserving the efficiency and rendering quality of 3DGS. Besides, our method competes favorably with leading surface reconstruction techniques while offering a more efficient learning process and much better rendering qualities. The code will be available at https://github.com/CVMI-Lab/3DGSR.