Depth correction-based underwater polarized light imaging method

Polarization imaging technology effectively improves the clarity of visible light images degraded by water scattering. Conventional methods typically estimate background light on a 2D plane, inherently neglecting its depth-dependent nature and suffering from severe low-frequency crosstalk between target and background. To overcome these limitations, this paper proposes a paradigm shift towards a depth-stratified estimation framework. We introduce a novel relative depth decoupling approach driven solely by polarization angle gradients, providing crucial 3D spatial priors without additional hardware. An adaptive depth binning strategy is then developed to construct an isolation mechanism that eliminates low-frequency target interference within discrete depth layers, ensuring high-fidelity background reconstruction even in structurally complex regions. Furthermore, by incorporating Malus' Law, the proposed approach effectively describes the non-uniform polarization states of diverse materials, breaking the conventional uniform-polarization assumption. Experimental results verify that our method significantly enhances contrast and detail restoration, exhibiting strong robustness in complex, high-turbidity underwater scenes.