Image reconstruction of underwater objects using regression-estimated remote Mueller matrix polarimetry.

The Mueller matrix provides a complete description of how a medium modifies the polarization state of light. Here, we present a polarimetric imaging approach that combines Mueller matrix analysis with regression modeling to reconstruct the air-equivalent image of submerged objects in water with reduced medium-induced distortion. By learning the relationship between polarization measurements acquired in air and through water, we estimate the Mueller matrix of the intervening medium without direct access to it. The estimated matrix is then used to compensate for polarization distortions and recover clearer images of underwater objects. This approach enables remote reconstruction of previously unseen objects and offers a practical strategy for underwater imaging, sensing, and operation in scattering environments.