Related Experiment Video
Updated: Jun 24, 2025
14:09
High-Throughput Total Internal Reflection Fluorescence and Direct Stochastic Optical Reconstruction Microscopy Using a Photonic Chip
Published on: November 16, 2019
6.9K
Time-gated imaging through dense fog via physics-driven Swin transformer
Optics Express
|June 11, 2024
Summary
This study introduces a novel physics-driven Swin Transformer method for clear imaging through dense fog. The technique effectively reconstructs targets even in severe fog conditions, outperforming existing methods.
Area of Science:
- Optics
- Computer Vision
- Artificial Intelligence
Background:
- Dense fog significantly degrades imaging quality due to severe light scattering and backscattering.
- Existing imaging methods struggle with reduced photon correlations, limiting target reconstruction in foggy environments.
Purpose of the Study:
- To develop an advanced imaging method capable of mitigating scattering effects in heterogeneous dense fog.
- To reconstruct target objects effectively despite significant attenuation of ballistic photons.
Main Methods:
- A physics-driven Swin Transformer model was developed, integrating Time-of-Flight (ToF) principles with Deep Learning.
- The method models the optical scatter imaging process to counteract fog-induced distortions.
Main Results:
- The proposed method demonstrated satisfactory performance in imaging targets obscured by dense fog.
- Successful reconstruction was achieved even at optical thicknesses up to 3.0, exceeding previous research limits.
- Quantitative metrics (PSNR, SSIM) confirm the cutting-edge performance of the method.
Conclusions:
- The physics-driven Swin Transformer method offers a significant advancement in imaging through dense fog.
- This approach holds promise for applications like autonomous driving and cosmic exploration requiring robust vision in adverse weather.
