发表机构
The University of Tokyo; Adelaide University; School of Artificial Intelligence (SAI) Shanghai Jiao Tong University; Hong Kong Polytechnic University（东京大学; 阿德莱德大学; 上海交通大学人工智能学院; 香港理工大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文针对黑暗环境下低光照RGB-NIR成像鲁棒性不足的问题，提出一种无需干净RGB监督、具备三维感知能力的神经模型，可融合含噪RGB与NIR线索恢复干净RGB图像，经合成与真实数据验证效果优越。
AI 中文摘要
低光照成像的鲁棒性仍是该领域面临的挑战。近期研究探索将近红外（NIR）与含噪RGB图像融合以实现更优的增强效果，但多数方法依赖精心整理的训练数据对，在不同场景下的鲁棒性有限。本文为RGB-NIR低光照成像提供了新视角，引入具备三维感知能力的神经建模方法。无需使用干净的RGB监督数据，即可优化出强大模型，在三维空间中隐式融合含噪程度极高的RGB观测结果与NIR线索，有效恢复干净的RGB图像。该模型省去了收集干净RGB数据的需求，可在不同噪声水平间泛化。在合成数据与真实数据上开展的大量评估表明其具有优越性。代码可访问：this https URL
英文摘要
Robust low-light imaging remains challenging for the community. Recent studies have explored fusing Near-Infrared (NIR) with noisy RGB to achieve improved enhancement, yet most methods depend on carefully curated training data pairs, with limited robustness under different scenarios. This paper offers a new perspective for RGB-NIR low-light imaging by incorporating 3D-aware neural modeling. Without using clean RGB supervision, a powerful model can be optimized to implicitly fuse extremely noisy RGB observations with NIR cues in 3D space, effectively recovering clean RGB images. The proposed model obviates the requirement for clean RGB data collection, generalizes across different noise levels. Extensive evaluations on synthetic and real data demonstrate its superiority. Codes available: https://github.com/MyNiuuu/3DarkFusion
CommentsACM Multimedia 2026, Codes and Models: https://github.com/MyNiuuu/3DarkFusion
