Iterative segmentation algorithm for enhancing the accuracy of biomimetic polarized light compass.

This article addresses the main challenge of poor environmental adaptability in bio-inspired polarized optical compasses, which leads to significant heading angle errors under complex weather and occlusion conditions. It presents a theoretical analysis and an analytical model of atmospheric multiple scattering effects on polarization patterns, validated by Monte Carlo simulations. To mitigate errors caused by asymmetric occlusions and model inaccuracies, the authors propose an iterative symmetric segmentation image processing algorithm that restores symmetry in polarization images, thereby reducing heading angle errors from 11.7° to 1.4° while maintaining computational efficiency suitable for high-frequency navigation. Experimental results under various weather conditions and building occlusions demonstrate that this algorithm achieves accuracy comparable to complex multiple scattering-based methods but with significantly faster processing times.