Underwater polarization information can provide reliable autonomous heading feedback for underwater scenes in the presence of satellite denial and multi-source interference. A biologically inspired highly robust polarization orientation strategy is proposed to address the degradation caused by weak polarization and noisy patterns in underwater environments with poor illumination, turbidity, and cloud recession. This approach initially develops a biomimetic algorithm that enhances weak polarization information, based on the increased perception mechanism of the visual neural pathway of syrphid fly under weak light conditions. After overcoming restricting polarization transmission, it is further optimized with a non-local sparse coding denoising part that takes into account the similarity in the detailed structure of the polarized image's effective zenith region. Relying on the stable ∞ symmetric distribution of the angle of polarization, the ±π/2 feature point distribution is traversed and assigned, and polarization information is quantitatively assigned, compensated, and regulated along the solar meridian region, excluding invalid polarized pixels. The experiments demonstrate that the methodology effectively reconstructs the symmetry of the partially damaged angle of polarization distribution, eliminates external interference, autonomously solves the fitted heading correction problem, and improves the environmental adaptability of the bionic polarization compass. The data from underwater experiments indicate the orientation accuracy: the error in weakly polarized patterns is within 2°, and the accuracy in noisy patterns can be increased by over 50%.