Research On Noise Reduction Algorithm Of Polarized Sensor

Polarized light navigation sensor can be used in the manufacture of core components of high-end guidance equipment such as aerospace because of its good concealment and no cumulative error.However,in the actual application process,due to the influence of environmental random noise,its detection light intensity is prone to fluctuation,leading to the decrease of guidance precision,which seriously affects its practical application value.Therefore,it is very important to study the noise property of polarized light navigation sensor and design the angle algorithm to reduce the noise.Firstly,in order to suppress the influence of environmental random noise on the accuracy of polarized light navigation sensor,this paper introduces Gaussian distribution theory and establishes a statistical solution model of polarization angle with physical properties.This model has high accuracy and stability,and can improve the accuracy of polarized light navigation sensor under certain interference environment.When the noise is large,the polarization angle error calculated by the traditional method and the statistical algorithm increases due to the limitation of the physical model.When the noise is moderate,by comparing the light intensity ratio and estimating the parameters,the statistical algorithm can find the optimal solution and improve the accuracy.When the noise is weak,due to the limitations of the algorithm itself,the comparison and optimization functions are weakened,and the statistical solution algorithm can only slightly improve the solution accuracy.At this time,the error is mainly determined by the system itself.Secondly,the light intensity noise model is established to avoid nonlinear noise model.In the statistical model,the light intensity ratio is used as a random variable.However,the light intensity ratio is obtained by dividing the transmitted light intensity,which is not suitable for linear hypothesis.Therefore,the light intensity noise is selected as the random variable for modeling.In addition,the theoretical angle accuracy is calculated by numerical calculation.According to the experiment,the average sample angle accuracy estimated by the light intensity noise model is roughly in line with the average sample angle accuracy measured by the experiment.Thirdly,to deal with random disturbances other than noise,the minimum information entropy algorithm is proposed to remove abnormal samples.In the real environment,there are random disturbances in addition to noise,which are characterized by short time and large amplitude.When a small number of abnormal pixels appear in the pixel array,the minimum information entropy algorithm can effectively remove the pixel marks deviating from the normal and remove them.However,when the abnormal pixel is much more than the normal pixel,the algorithm will take the abnormal value as the normal value and remove the normal pixel.However,even with this shortcoming,the minimum information entropy algorithm still has a good universality,because it is independent of scale in the calculation of information entropy,and can be used to remove a wide range of sample abnormal points.Finally,an experimental platform of polarized light sensor based on STM32 and CMOS is designed and built.The hardware structure of the sensor is designed,and the program design and implementation of the lower computer and the upper computer are completed.Through experiments,the above model and algorithm are verified.