PEU-Net: a Stokes-constrained lightweight framework for real-time polarization dehazing
Ziheng Shang, Junmeng Han, Long Jin, Yushi Jin, Yongji Yu, Yuan Dong · Chinese Optics Letters · 2025
Natural fog and industrial smoke severely degrade optical imaging visibility, undermining real-time applications like autonomous driving and unmanned aerial vehicle (UAV) surveillance.This makes haze removal a critical research challenge.This study proposes a polarimetric efficient U-Net (PEU-Net), a lightweight Stokes-constrained framework featuring a symmetrical U-Net encoder-decoder with optimized baseline channels, an efficient channel attention (ECA) module for dynamic channel calibration, and a Stokes-based loss function to suppress nonlinear noise.Ablation studies validate 1.3 dB/2.29 dB gains from ECA and Stokes loss, respectively.The framework achieves real-time processing at 33.98 frame per second (FPS) for 1024 × 1024 inputs and attains 23.04 dB peak signal-to-noise ratio (PSNR) and 0.821 structural similarity index measure (SSIM) on controlled fog chamber datasets, outperforming state-of-the-art methods in accuracy and efficiency.This research provides an efficient solution for real-time polarization-based dehazing tasks and offers, to our knowledge, and offers novel insights into the design of lightweight physics-data jointly driven models.