Study On Orientation Of Atmospheric Polarized Light In Complex Weather

As an important part of unmanned motion platform,navigation and orientation technology has attracted much attention.In order to solve the navigation and orientation problems of long endurance and anti-interference without relying on satellites,navigation and orientation technology based on atmospheric polarized light has developed rapidly.However applied in ground or low altitude atmosphere polarized light navigation technology,the directional accuracy besides affected by geographical location and time,still heavily dependent on the detection of the weather,in this paper,on the basis of the current atmospheric polarization orientation technology research,proposed a method of atmospheric polarized orientation under complicated weather,and through a series of experiments verify the feasibility of the proposed method,The relevant work is as follows:Firstly,the orientation technology of atmospheric polarized light is systematically explained,and the orientation algorithm based on zenith ROI region and fitting solar meridian are introduced.An orientation system based on atmospheric polarized light is built.SONY IMX250 MZR is used as a polarization detection chip.The program is written in Linux environment and verified by experiments.Secondly,the noise analysis of polarization images under complex weather conditions is carried out to determine the noise mode of polarization Angle under complex weather conditions,and the data set under sunny,overcast,smog and sandstorm weather is built,and the classification of different weather is carried out by using PCANet network.Then,aiming at the high precision orientation requirements,a polarization orientation method of weather weighted sparse coding is proposed by combining weather classification and trilateral weighted sparse coding,and experimental verification is carried out under complex weather conditions.Experimental results show that the polarization orientation algorithm based on weather-weighted sparse coding can effectively improve the orientation accuracy of atmospheric polarized light under complex weather conditions.Finally,aiming at the application conditions of real-time,a fast direction algorithm based on Deam Net network was proposed,and experiments of remote control car,automobile and six-rotor UAV were carried out to verify the feasibility of the proposed algorithm.