Turbid underwater polarization imaging method using a single intensity-only image

Polarization imaging separates scattered and target light through polarization analysis, enabling clear imaging in turbid underwater environments. However, existing polarization imaging methods strictly rely on multiple captured images or specialized hardware such as division of focal plane (DoFP) polarization cameras, which entail high hardware complexity, cost and light loss. To overcome these limitations, this paper proposes an intensity-only image polarization restoration method that eliminates the need for any polarization hardware. Unlike traditional methods requiring physical polarization acquisition, our approach constructs valid pseudo-orthogonal polarization image pair directly from a single intensity input. Specifically, the intensity input is first decomposed into polarized and unpolarized components, which are then recombined to generate a pair of pseudo-orthogonal polarization images. From this constructed pair, the backscattered intensity is estimated via pyramid decomposition, flat-region detection and pyramid reconstruction, and then integrated into a polarization model for clear scene restoration. Experimental results demonstrate that our method can generate a pair of valid orthogonal polarization images from a single intensity-only image, and when coupled with the polarization-based restoration model, it enables effective scene recovery across various underwater environments.