Research On High-precision Deformation Inversion Method Of Ground-based SAR Based On Long-term Sequence Observation Error Correction

Ground-based synthetic aperture radar(SAR)differential interferometry is a technique based on synthetic aperture radar to monitor slight deformations.The ground-based SAR system has the characteristics of all-weather,all-weather,non-contact,large observation area,and unlimited light.At the same time,because the platform is statically placed on the ground,compared with traditional spaceborne and airborne radars,the Ground-based SAR systems have the advantages of short observation periods and real-time fast response to short-term changes in deformation regions.In currently published literature,this technology has been applied to surface mine monitoring,dam safety monitoring,landslide rescue and rescue.This article summarizes the research status of ground-based SAR systems,the status of related article publications,the research status of deformation measurement technology,and the status of ground-based SAR system applications.The working mode,signal system and imaging algorithm of ground-based SAR system are introduced,the general flow of deformation inversion algorithm is described,and the sources of deformation error of ground-based SAR system are analyzed.Because ground-based SAR has an observation phase error due to environmental changes during long-term external field observations,the observation phase error will form a deformation error,which will affect the system monitoring accuracy.Therefore,in this paper,aiming at the problem that environmental changes affect the monitoring accuracy,it is proposed to carry out a research on the high-precision deformation inversion method of ground-based SAR based on long-term serial observation error correction.For the single threshold only emphasizes a certain characteristic of the PS point,the double threshold method does not consider the problem of phase instability.This paper proposes a three threshold algorithm for the combination of coherence coefficient,amplitude dispersion,and phase error.This method is based on the double threshold algorithm On the basis of fully considering the influence of phase error information on the selection of permanent scatterer(Permanent Scatterer,PS)points,we will further investigate the fine amount of phase error of pixel stability as a method of screening PS points,by setting a reasonable phase error threshold The PS points were screened,and the accuracy of the algorithm was verified by experiments.Then,when using the threshold method to filter permanent scatterers,the process of manually debugging multiple thresholds is complicated and prone to unreasonable threshold setting.A multi-threshold optimization algorithm is proposed.This algorithm automatically determines the threshold for screening permanent scatterers for ground-based radar,Can reduce the unreasonable phenomenon of manually setting the threshold,save the process of complicated threshold debugging after replacing the monitoring scene,and the automatically determined threshold screening result achieves a better effect than the manually adjusted threshold.Next,in this paper,it is difficult to realize the use of different size thresholds for different observation areas when facing large scenes.Combining machine learning methods with traditional ground-based SAR deformation inversion algorithms,a Transformer model is used to classify radar pixels Method,this method first preprocesses the ground-based SAR data,and then enters the data into the Transformer model.Finally,the model selects pixels that are suitable for deformation inversion.The improved threshold method can provide better data sets for this method,and the threshold method can be compared with this method to verify the experimental results.Experimental results prove that the method of Transformer classifying pixels is feasible.