DelowlightSplat: 面向低光照3D场景重建的前馈高斯泼溅
DelowlightSplat: Feed-Forward Gaussian Splatting for Lowlight 3D Scene Reconstruction
- Hangzhou Dianzi University（杭州电子科技大学）
- Zhuhai College of Science（珠海科技学院）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
提出DelowlightSplat，一种低光照感知的前馈高斯泼溅框架，通过轻量级低光照适配器和成本体积多视图推理，从稀疏有噪声图像中直接预测干净3D高斯，实现高质量新视角合成。
AI中文摘要：
从稀疏有姿态图像进行新视角合成和3D重建是机器人和AR/VR的核心。然而，前馈3D高斯重建在低光照下因噪声、颜色偏移和不可靠对应而失败。我们提出DelowlightSplat，一种低光照感知的前馈高斯泼溅框架，用于干净的新视角渲染。我们通过仅退化上下文视图同时保持目标视图干净，构建了一个可控的多视图低光照基准。我们引入轻量级低光照适配器进行残差增强以提高可匹配性，并将其与基于成本体积的多视图推理相结合，直接预测干净的3D高斯。实验表明，DelowlightSplat在低光照条件下显著优于先前的前馈方法和两阶段流水线。
英文摘要：
Novel-view synthesis and 3D reconstruction from sparse posed images are central to robotics and AR/VR. Yet, feed-forward 3D Gaussian reconstruction fails under lowlight due to noise, color shifts, and unreliable correspondence. We propose DelowlightSplat, a lowlight-aware feed-forward Gaussian splatting framework for clean novel-view rendering. We build a controllable multi-view lowlight benchmark by degrading only context views while keeping target views clean. We introduce a lightweight Lowlight Adapter for residual enhancement to improve matchability, and couple it with cost-volume-based multi-view inference to directly predict clean 3D Gaussians. Experiments show that DelowlightSplat significantly outperforms previous feed-forward method and two-stage pipeline under lowlight conditions.
