A highly robust polarization orientation method based on antisymmetry of skylight polarization pattern

Abstract

The bionic polarized light compass has characteristics of immunity to interference from electromagnetic fields and zero error accumulation over time, and thus has good application prospects in the autonomous navigation field under conditions of Global Positioning System rejection or failure. With the aim of addressing the problem that the polarization orientation accuracy of the bionic polarized light compass can be degraded seriously by the presence of large-area occlusions such as clouds and buildings during the actual application process, this paper proposes a highly robust orientation method under the limited effective polarization area condition based on the antisymmetry of the skylight polarization pattern. The orientation algorithm first uses the UNet convolutional neural network to perform binary segmentation of the polarized image and then separates and removes the thick cloud layer/building occlusion area from the image; Then, a new solar meridian fitting method is proposed, which makes use of the characteristic of antisymmetric distribution of polarization angle along the solar meridian in atmospheric polarization mode to fit the solar meridian, and then completes the orientation. Experimental results show that the method proposed in this paper can effectively solve the problem of polarization orientation accuracy degradation caused by harsh environments, including clouds, building occlusions, and cloudy days, and can thus improve the robustness of bionic polarized light compass orientation effectively in complex environments.