Influence Analysis And Calibration Of SAR Polarimetric Measurement Errors

Synthesis Aperture Radar(SAR)is a high-resolution radar that enables continuous ground observations throughout the day and all weather.The introduction of polarimetric information not only provides radar multi-channel data support,but also facilitates the acquisition of more complete physical property information of radar targets,which is of great significance.However,due to the increase of observation dimension and the increase of system complexity,the polarimetric data of SAR measurement will be affected by system errors such as radar antenna crosstalk and radar electromagnetic wave propagation environment,which will not accurately reflect the true polarimetric scattering characteristics of the target,and affects subsequent polarimetric applications.This paper focuses on the effects of SAR polarimetric measurement errors on the main polarimetric applications(polarimetric detection,scattering characteristics extraction based on polarimetric decomposition)and how to accurately calibrate the polarimetric measurement errors.Firstly,the influence of SAR polarimetric measurement errors on polarimetric detection performance is analyzed.The degree of distortion of the polarimetric data under the polarimetric error is analyzed,and the model of the radar echo is established.For the classical polarimetric detectors(OPD,PMF,SD,PWF and PMS),the radar has high resolution and low and medium resolution,and the detection performance in the case is analyzed.Considering polarimetric crosstalk,polarimetric channel imbalance and Faraday rotation three kinds of polarimetric measurement errors,a detection performance evaluation method based on optimal detection offset coefficient is proposed,and the change of probability of detection is analyzed,which provides support for error control during the polarimetric measurement process and polarimetric detection of radar targets.The second part analyzes the influence of SAR polarimetric measurement errors on scattering characteristics extraction based on polarimetric decomposition,which is divided into two types: coherent target decomposition(Pauli decomposition,Krogager decomposition)and partial coherent target decomposition(Cloude-Pottier decomposition).Firstly,the principle and significance of polarimetric decomposition are briefly introduced.Next,the effects of polarimetric crosstalk,polarimetric channel imbalance and Faraday rotation on the measurement errors are simulated.For coherent target decomposition,the polarimetric measurement errors is analyzed from the normalized error histogram of the polarimetric decomposition feature vector,the intensity variation histogram of each element of the feature vector,and the deviation gray image before and after the error of the single feature vector element.For the partial coherent target decomposition,the variation trend of the scattering mechanism of the typical object target center and the deviation ratio of the classification result are used to analyze the influence of the polarimetric measurement errors,which provides support for error extraction based on polarimetric decomposition and error control during polarimetric measurement.Finally,the accurate calibration of SAR polarimetric measurement errors is studied.It is divided into two types: polarimetric measurement error of SAR system(polarimetric channel imbalance,polarimetric crosstalk)and polarimetric measurement error(Faraday rotation)caused by SAR electromagnetic wave propagation environment.For the calibration problem of polarimetric measurement error of SAR system,a polarimetric calibration algorithm based on particle swarm optimization is proposed.The simulation experiment proves the effectiveness of particle swarm optimization for polarimetric calibration,and the algorithm has a more stable and accurate estimation.For the correction of Faraday rotation,a new Faraday rotation angle estimation method based on polarimetric covariance matrix is proposed.Compared with the classical estimation method,the estimation method utilizes more raw data information.,and has a higher noise margin.