GaussianRoom: Improving 3D Gaussian Splatting with SDF Guidance and Monocular Cues for Indoor Scene Reconstruction

Haodong Xiang,Xinghui Li,Xiansong Lai,Wanting Zhang,Zhichao Liao,Kai Cheng,Xueping Liu

2024-05-30

Abstract:Recently, 3D Gaussian Splatting(3DGS) has revolutionized neural rendering with its high-quality rendering and real-time speed. However, when it comes to indoor scenes with a significant number of textureless areas, 3DGS yields incomplete and noisy reconstruction results due to the poor initialization of the point cloud and under-constrained optimization. Inspired by the continuity of signed distance field (SDF), which naturally has advantages in modeling surfaces, we present a unified optimizing framework integrating neural SDF with 3DGS. This framework incorporates a learnable neural SDF field to guide the densification and pruning of Gaussians, enabling Gaussians to accurately model scenes even with poor initialized point clouds. At the same time, the geometry represented by Gaussians improves the efficiency of the SDF field by piloting its point sampling. Additionally, we regularize the optimization with normal and edge priors to eliminate geometry ambiguity in textureless areas and improve the details. Extensive experiments in ScanNet and ScanNet++ show that our method achieves state-of-the-art performance in both surface reconstruction and novel view synthesis.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

### Problems Addressed by the Paper

This paper aims to address issues in indoor scene reconstruction, particularly the incomplete and noisy reconstruction results in texture-less areas. Specifically, the paper proposes a new method called **GaussianRoom**, which combines Neural Signed Distance Field (SDF) with 3D Gaussian Splatting (3DGS) to improve the geometric reconstruction of indoor scenes while maintaining efficient rendering performance.

#### Main Contributions

1. **Unified Optimization Framework**:
- Proposes a unified optimization framework that combines neural SDF with 3DGS.
- Designs an SDF-guided primitive distribution strategy to guide the densification and pruning of Gaussian points, enabling accurate scene modeling even with poor initial point cloud quality.
- Utilizes the geometric structure represented by 3DGS to enhance the sampling efficiency of the SDF field.

2. **Monocular Geometric Cues**:
- Introduces monocular normal priors and edge priors to eliminate geometric ambiguities in texture-less areas and improve details.

3. **Experimental Validation**:
- Extensive experiments on the ScanNet and ScanNet++ datasets demonstrate that the proposed method achieves state-of-the-art performance in both surface reconstruction and novel view synthesis.

Through the aforementioned methods, GaussianRoom significantly enhances the quality of indoor scene reconstruction and rendering effects, especially excelling in handling texture-less areas.