NeRT: Implicit Neural Representations for Unsupervised Atmospheric Turbulence Mitigation
The atmospheric turbulence mitigation problem has emerged as a challenging inverse problem in the communities of computer vision and optics. However, current methods either rely heavily on the quality of the training dataset or fail to generalize over various scenarios, such as static scenes, dynamic scenes, and text reconstructions. We propose a novel implicit neural representation for unsupervised atmospheric turbulence mitigation (NeRT). NeRT leverages the implicit neural representations and the physically correct tilt-then-blur turbulence model to reconstruct the clean and undistorted image, given only dozens of distorted images. Further, we show that NeRT outperforms the state-of-the-art through various qualitative and quantitative evaluations. Lastly, we incorporate NeRT into continuously captured video sequences and demonstrate 48 times speedup.