Research On Polarimetric Calibration Of SAR Images Using Distributed Targets

Polarimetric SAR(PolSAR)is an important technology to achieve remote earthsurface observation.It can acquire images on wide range of earth surface in day or night,and in all weather conditions.It can provide rich information for ground target interpretation,and has been widely used in surveying,oceanography,forestry,agriculture,military,geological,archaeological and other fields.As PolSAR is useful in observation,many countries including china have been developing new kinds of PolSAR systems in recent years.These systems provide large amount of data for research and applications in remote sensing.Meanwhile,they require new and better polarimetric calibration(Pol CAL)methods.During the observation,the system inevitably generates polarimetric errors,for instance the antenna crosstalk,the channel imbalance,the polarimetric phase imbalance,and the system noise.Therefore,the observed images need Pol CAL processing before they can be used for any quantitative research or applications.The current polarimetric calibration research still has many problems.For example,the Pol CAL needs large work of calibrator deployment,the extraction of referenced targets for Pol CAL is not accurate enough,the methods based on distributed target will underestimate crosstalk error,and the evaluating methods of polarimetric quality are not objective enough.After analyzing the scattering and statistic properties of distributed targets,this dissertation provided several methods to improve the quality of PolSAR images.This dissertation proposes a method to extract Pol CAL referenced target accurately,and proposed a method to solve polarimetric errors using manmade calibrators and distributed targets,and it also provided a solution to assess the polarimetric quality of SAR images systematically.At present,there are mainly two kinds of Pol CAL methods.Methods based on distributed targets calibrate polarimetric errors using reflection symmetry or scattering reciprocity assumption.These methods can reduce the dependence on man-made calibrators,making them suitable for the calibration of airborne SAR images.However,these methods will underestimate the crosstalk error and the extraction of referenced distributed targets is still inaccurate.Another kind of methods calibrate all the polarimetric errors using man-made calibrators.These methods can calculate polarimetric errors accurately and objectively,but they require a lot of work on the construction of calibration field in the early stage,making them more suitable for the calibration of spaceborne SAR images.Besides,these methods have not considered the influence of system noise in the calibration.This dissertation presents the research on these two kinds of methods,and the specific work is within the following aspects:(1)This dissertation presents a systematic analysis on the signal propagation and the error sources during the observation of PolSAR system.It analyzed the influence of crosstalk,channel imbalance,and the phase imbalance on the scattering of target using polarimetric distortion model and trihedral corner reflector.The framework of polarimetric calibration of SAR images is also introduced.The Pol CAL methods are divided into three parts,the methods based on calibrators,the methods using both calibrators and distributed targets,and the methods based on distributed targets.The basic principles,research trends,advantage and problems of these Pol CAL methods are summarized and analyzed.(2)In the aspect of Pol CAL that based on distributed targets,the inaccurate extraction of referenced targets for Pol CAL may limit the calibration accuracy.This dissertation proposed an automatic method for Pol CAL referenced targets extraction,using helix scattering ratio feature.The helix scattering ratio was constructed basing on the polarimetric error model.The tolerance of the feature to different types of polarization error is analyzed.A new threshold segmentation algorithm is established using the statistical characteristics of this feature.This method can extracte reflection symmetry targets automatically.The experimental results show that the proposed method has high tolerance to polarimetric error,and can improve the calibration effect for algorithms that based on the distributed targets,and improve the polarimetric calibration accuracy finally.(3)In the aspect of polarimetric error solution,this dissertation proposed a Pol CAL method using a combination of calibrators and distributed targets.The method was settled up basing on the study of the relationship of system noise and polarimetric errors,and the study of their influence on the image quality.It is more suitable for the calibration of spaceborne SAR images.The method uses polarimetric active calibrators to solve the crosstalk error,so it can avoid the problem of crosstalk underestimating that arisen from methods that based on distributed targets.The channel imbalance errors are solved using the scattering reciprocity assumption of distributed targets,and the interference of the system noise to calibration can be reduced.The experiment is taken with GF-3 satellite imagery.And it shows that the method can ensure the scattering characteristics of ground objects after calibration and it can improve the polarimetric quality of the image.(4)In the aspect of quality evaluation and improvement of PolSAR images,this dissertation firstly introduced the principle of the flight experiment design on Pol CAL,and the dissertation also concluded the problems that should be paid attention to,which includes the selection of the calibration field,the layout and placement of the calibrators,the flight conditions,etc.This dissertation proposed a systematic scheme on polarimetric quality evaluation.Images of GF-3 satellite and domestic airborne SAR system were evaluated using this scheme,and the characteristics of methods in this scheme were analyzed and compared.The experimental result shows that the scheme can reveal the quality of the images comprehensively and objectively.The images were also re-calibrated by the method proposed in this dissertation.The experimental result shows that the calibrated images are correct,and the calibration process can reduce the polarimetric residual errors and improve the image quality.