Optimization Reserch On Airborne SAR Polarimetric Calibration Methods

Fully polarimetric SAR data are useful for understanding the polarization reflection properties of the observed objects in many applications such as geographical condition monitoring and environmental change study. Airborne SAR system not only can perform to the ground observation task independently, but also can be used as the basis and experimental platform on the study of the spaceborne SAR system and it is important to study the polarization of SAR. With the development of the new system, the rationality and the scientificity of the mass data has been challenged. Due to the inherent defects of the antenna system, the quality of the polarimetric data is affected and the acquisition of the correct information is disturbed. At the same time, people's demand for higher quality data is also increasing, which makes it more important to study the polarimetric calibration (PolCAL). In order to further utilize fully polarimetric data, polarimetric calibration is a fundamental pre-processing step, which focuses on the removal of the radar antenna system distortion effect. For polarimetric distortion phenomena, a lot of research work has been carried out. Compared to earlier researches, the current researches require more stable and accurate polarimetric calibration method. A complete polarimetric calibration method includes the use of the calibration target and the algorithm to calculate parameters. The optimization research on the polarimetric calibration method is divided into two directions:one is the refinement of the calibration target for obtaining the prior information, another one is the stability and accuracy of the optimization calculation. In this paper, the two directions are further studied. The main contents of the research are as follows:(1) In the actual polarimetric measurements, a lot of distributed targets may often no longer meet the principle of reciprocity, due to the non-single station condition, Faraday rotation effect, nonlinear media. The reciprocity judgement of the calibration target, especially for under polarimetric distortion, it is the lack of sufficient research. To solve this problem, we propose an optimization method of reciprocity judgement to select calibration samples. We derive the expression of reciprocity angle and its monotonicity characteristics. Using the criterion of the smaller value, we refer to the reciprocity angle of some samples' observation value for determining reciprocity of the targets. After the initial judgement of the outstanding reasonable samples, coherence test in sliding window is finally settled for selecting these reciprocal samples. Then it analyzes the impact of the optimization method. Through the weak terrain scattering in RGB image, the distortion of polarimetric feature response, the change about reciprocity angle, it verifies the validity and reliability of the optimization method. From the cross polarization energy variation, clustering center changes to the precision on statistics analysis, it shows the importance of target's reciprocity in unsupervised classification of ground objects.(2)For airborne synthetic aperture radar (SAR) polarimetric calibration based on distributed targets, it is important to ensure the removal of both the polarimetric distortion and terrain slope effect. Our solution to this problem, a new technique to PolCAL using a priori asymmetry information of distributed targets from the data fusion is proposed and evaluated in this paper. DEM generated from interferometry in X-band can provide a priori asymmetry information (POAS) for fully polarimetric P-band data. The synchronicity and the manual registration is capable to ensure high accuracy. Using the POAS, the calibration model relying on distributed targets is determined and segmented into two parts to solve. The crosstalk and x-pol channel imbalance are iteratively determined and the proposed method is closer to the results of using CRs than other methods. The displayed polarimetric signatures from several CRs were shown that the proposed technique could be similar to the using CRs method. And this proposed technique maintained high SNR and did not damage randomness of several typical distributed targets. The comparison studies on two datasets by using the proposed technique and the first calibration parameters were conducted and showed that the proposed technique have better performance for POAS estimation. In long-term operation, the proposed technique is capable to monitoring system condition and calibration parameters can be corrected. Similar to CASMSAR system, the other multi-band airborne SAR systems with fully polarimetry and interferometry can also apply this technique.(3) The problems in the existing polarimetric calibration parameters calculation are not stable and low accuracy. Stability means the algorithm can solve almost all the specific polarimetric distortion. Accuracy is the exact degree of the parameters value. To solve the problems, indeterminate equations, positive definite equations and over-determined equations are proposed. Using several mathematical optimization algorithms, the corresponding calibration optimization algorithms are established. And the reasonable and effective algorithms'processes are designed. After confirming the relationship between the calibration targets and the equation, the basic principle of the optimization is given, which is to avoid the division formula be zero. In this paper, a new algorithm for the stability and accuracy of the polarimetric calibration optimization is proposed. And this Levenberg-Marquardt (LM) algorithm is based on the least square principle. Using the LM optimization algorithm, corner reflectors can be reduced to two from at least three. The accuracy of the distributed targets calculation is not as good as that of the point target, but the stability and the accuracy of the algorithm are still guaranteed. In the calibration site experiment, deploying a trihedral corner reflector and a 45 degree dihedral corner reflector, using LM optimization algorithm, the validity of the algorithm is also proved.