Heading Angle Measurement Algorithm for Polarized Light Based on a Berry Model Optimized by PSO

Polarized light navigation enables heading angle measurement by leveraging atmospheric polarization patterns, offering advantages such as error-free accumulation and resistance to interference, making it suitable for GNSS-denied environments. However, traditional optimization algorithms applied to the Berry model often fail to achieve sufficient accuracy, reducing robustness under complex atmospheric conditions. To address this, this paper proposes an improved heading estimation method: K-means clustering is first employed to segment sky polarization images and eliminate occlusion interference; then, Particle Swarm Optimization (PSO) is used to optimize the Berry model parameters, enhancing its ability to represent real atmospheric polarization patterns. Experimental results show that, compared with the Sequential Least Squares Quadratic Programming (SLSQP) method, the proposed method significantly improves heading angle accuracy, reducing the root mean square error by 59.43% and the mean error by 19.38%, demonstrating its effectiveness in complex environments.