Dual-Dimensional Gaussian Splatting Integrating 2D and 3D Gaussians for Surface Reconstruction
Three-Dimensional Gaussian Splatting (3DGS) has revolutionized novel-view synthesis, enabling real-time rendering of high-quality scenes. Two-Dimensional Gaussian Splatting (2DGS) improves geometric accuracy by replacing 3D Gaussians with flat 2D Gaussians. However, the flat nature of 2D Gaussians r...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-06-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/15/12/6769 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849433440248659968 |
|---|---|
| author | Jichan Park Jae-Won Suh Yuseok Ban |
| author_facet | Jichan Park Jae-Won Suh Yuseok Ban |
| author_sort | Jichan Park |
| collection | DOAJ |
| description | Three-Dimensional Gaussian Splatting (3DGS) has revolutionized novel-view synthesis, enabling real-time rendering of high-quality scenes. Two-Dimensional Gaussian Splatting (2DGS) improves geometric accuracy by replacing 3D Gaussians with flat 2D Gaussians. However, the flat nature of 2D Gaussians reduces mesh quality on volumetric surfaces and results in over-smoothed reconstruction. To address this, we propose Dual-Dimensional Gaussian Splatting (DDGS), which integrates both 2D and 3D Gaussians. First, we generalize the homogeneous transformation matrix based on 2DGS to initialize all Gaussians in 3D. Subsequently, during training, we selectively convert Gaussians into 2D representations based on their scale. This approach leverages the complementary strengths of 2D and 3D Gaussians, resulting in more accurate surface reconstruction across both flat and volumetric regions. Additionally, to mitigate over-smoothing, we introduce gradient-based regularization terms. Quantitative evaluations on the DTU and TnT datasets demonstrate that DDGS consistently outperforms prior methods, including 3DGS, SuGaR, and 2DGS, achieving the best Chamfer Distance and F1 score across a wide range of scenes. |
| format | Article |
| id | doaj-art-a16863fc0e934386afcbff2ee35d7911 |
| institution | Kabale University |
| issn | 2076-3417 |
| language | English |
| publishDate | 2025-06-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj-art-a16863fc0e934386afcbff2ee35d79112025-08-20T03:27:02ZengMDPI AGApplied Sciences2076-34172025-06-011512676910.3390/app15126769Dual-Dimensional Gaussian Splatting Integrating 2D and 3D Gaussians for Surface ReconstructionJichan Park0Jae-Won Suh1Yuseok Ban2Department of Electronics Engineering, Chungbuk National University, 1 Chungdae-ro, Seowon-gu, Cheongju 28644, Republic of KoreaDepartment of Electronics Engineering, Chungbuk National University, 1 Chungdae-ro, Seowon-gu, Cheongju 28644, Republic of KoreaDepartment of Electronics Engineering, Chungbuk National University, 1 Chungdae-ro, Seowon-gu, Cheongju 28644, Republic of KoreaThree-Dimensional Gaussian Splatting (3DGS) has revolutionized novel-view synthesis, enabling real-time rendering of high-quality scenes. Two-Dimensional Gaussian Splatting (2DGS) improves geometric accuracy by replacing 3D Gaussians with flat 2D Gaussians. However, the flat nature of 2D Gaussians reduces mesh quality on volumetric surfaces and results in over-smoothed reconstruction. To address this, we propose Dual-Dimensional Gaussian Splatting (DDGS), which integrates both 2D and 3D Gaussians. First, we generalize the homogeneous transformation matrix based on 2DGS to initialize all Gaussians in 3D. Subsequently, during training, we selectively convert Gaussians into 2D representations based on their scale. This approach leverages the complementary strengths of 2D and 3D Gaussians, resulting in more accurate surface reconstruction across both flat and volumetric regions. Additionally, to mitigate over-smoothing, we introduce gradient-based regularization terms. Quantitative evaluations on the DTU and TnT datasets demonstrate that DDGS consistently outperforms prior methods, including 3DGS, SuGaR, and 2DGS, achieving the best Chamfer Distance and F1 score across a wide range of scenes.https://www.mdpi.com/2076-3417/15/12/6769surface reconstructionview synthesisGaussian splatting |
| spellingShingle | Jichan Park Jae-Won Suh Yuseok Ban Dual-Dimensional Gaussian Splatting Integrating 2D and 3D Gaussians for Surface Reconstruction Applied Sciences surface reconstruction view synthesis Gaussian splatting |
| title | Dual-Dimensional Gaussian Splatting Integrating 2D and 3D Gaussians for Surface Reconstruction |
| title_full | Dual-Dimensional Gaussian Splatting Integrating 2D and 3D Gaussians for Surface Reconstruction |
| title_fullStr | Dual-Dimensional Gaussian Splatting Integrating 2D and 3D Gaussians for Surface Reconstruction |
| title_full_unstemmed | Dual-Dimensional Gaussian Splatting Integrating 2D and 3D Gaussians for Surface Reconstruction |
| title_short | Dual-Dimensional Gaussian Splatting Integrating 2D and 3D Gaussians for Surface Reconstruction |
| title_sort | dual dimensional gaussian splatting integrating 2d and 3d gaussians for surface reconstruction |
| topic | surface reconstruction view synthesis Gaussian splatting |
| url | https://www.mdpi.com/2076-3417/15/12/6769 |
| work_keys_str_mv | AT jichanpark dualdimensionalgaussiansplattingintegrating2dand3dgaussiansforsurfacereconstruction AT jaewonsuh dualdimensionalgaussiansplattingintegrating2dand3dgaussiansforsurfacereconstruction AT yuseokban dualdimensionalgaussiansplattingintegrating2dand3dgaussiansforsurfacereconstruction |