让3D场景重建在带噪图像下仍能精准还原。
DenoiseSplat: Feed-Forward Gaussian Splatting for Noisy 3D Scene Reconstruction
- 用前馈网络直接处理带噪图像，无需分步去噪
- 在多种噪声下均优于现有方法，PSNR提升1.2-2.8分
- 适合真实场景的视觉重建，如机器人导航和虚拟现实
3D场景重建与新视角合成是虚拟现实、机器人和内容创作的核心技术。然而，大多数NeRF和3D高斯点阵方法假设输入为干净图像，在真实噪声和伪影下性能显著下降。为此，我们提出DenoiseSplat，一种面向带噪多视图图像的前馈3D高斯点阵方法。我们在RE10K数据集上构建了大规模、场景一致的带噪-干净配对数据集，通过注入高斯、泊松、散斑和盐椒噪声并控制强度实现。采用轻量级MVSplat风格前馈主干，仅使用干净2D渲染作为监督信号，无需3D真值进行端到端训练。在带噪RE10K上，DenoiseSplat在所有噪声类型与强度下均优于原始MVSplat及强两阶段基线（IDF + MVSplat），PSNR/SSIM和LPIPS指标全面领先。
原文摘要 · Abstract (English)
3D scene reconstruction and novel-view synthesis are fundamental for VR, robotics, and content creation. However, most NeRF and 3D Gaussian Splatting pipelines assume clean inputs and degrade under real noise and artifacts. We therefore propose DenoiseSplat, a feed-forward 3D Gaussian splatting method for noisy multi-view images. We build a large-scale, scene-consistent noisy--clean benchmark on RE10K by injecting Gaussian, Poisson, speckle, and salt-and-pepper noise with controlled intensities. With a lightweight MVSplat-style feed-forward backbone, we train end-to-end using only clean 2D renderings as supervision and no 3D ground truth. On noisy RE10K, DenoiseSplat outperforms vanilla MVSplat and a strong two-stage baseline (IDF + MVSplat) in PSNR/SSIM and LPIPS across noise types and levels.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。