Structure-Aware Consistency Priors for Shape from Polarization in Complex Media

Recovering surface normals from single-view polarization images in complex media remains challenging. This paper focuses on ice as a representative complex medium, where intricate light–matter interactions lead to a nonlinear mapping between polarization observations and surface normals. To address this, a structure-aware polarization prior based on autocorrelation functions is proposed to capture the local spatial consistency of AoLP. Building on this, a dual-branch network (IceSfP) is designed to integrate raw polarization features with priors via cross-modal attention and multi-scale feature fusion, enabling accurate surface normal estimation under complex media conditions. To evaluate the method, the first real-world ice SfP dataset is constructed. Experimental results show that the method outperforms existing approaches across all metrics, achieving a MAE of 16.01$^\circ$, which is 2.74$^\circ$ lower than the second-best method. The framework provides a generalizable solution for high-precision geometric perception in complex media.

Icy surfaces are hard to measure accurately because light scatters and bends unpredictably inside the ice. This makes it difficult for existing 3D imaging techniques to capture the true shape of ice, which is important for tasks like detecting icy roads, monitoring polar regions, and studying the environment. We developed a new method that uses the way ice reflects polarized light to figure out its shape. Our technique can automatically tell which parts of the ice give reliable signals and which parts are distorted by internal scattering. By combining these signals with physical models and a specially designed neural network, we can reconstruct high-precision 3D models of ice surfaces. To help other researchers, we also created the first dataset of 3D ice surfaces captured with polarization imaging. This approach is low-cost and passive, allowing machines to “see” ice reliably in challenging conditions, which could improve safety for autonomous vehicles and support ice and snow hazard monitoring.