Efficient Real-World Dehazing via Physics-Inspired Global-Local Decoupling
PGL-Net is a lightweight framework for real-world image dehazing that achieves state-of-the-art results with significantly reduced computational cost. It decouples dehazing into global distribution rectification using Physics-Inspired Affine Fusion (PAF) and local structural refinement via Degradation-Aware Modulation (DAM). This approach embeds physical priors to improve quality while drastically cutting inference time, making it suitable for real-time edge deployment.