Research On Key Issues Of Spaceborne Multi-angle SAR Imaging And Geometric And Scattering Information Extraction

Synthetic Aperture Radar(SAR)is an active microwave remote-sensor,featuring all-weather,all-weather and high resolution.It plays an important role in the field of remote sensing.Multi-angle SAR is an imaging mode that utilizes platform maneuvering and antenna beam steering to continuously observe the target area from different azimuth angles.It can provide scattering information,motion information and geometric information of the target from multi-angle observation and improve the efficiency of single SAR imaging.Conventional SAR satellites generally do not adopt azimuth multi-angle observation,and the azimuth observation angle capability is limited.For example,the advanced Terra SAR-X satellite can only provide ±2.2° maximum azimuth observation angle,while the azimuth observation angle variation of multi-angle SAR can generally reach ±30° or even higher.In the process of multi-angle observation,the azimuth Angle varies greatly,the range direction and azimuth direction of SAR echo signal are deeply coupled,which makes the SAR echo complex.Due to the long synthetic aperture time,it is difficult to model the relative geometric relationship between satellite and ground,and the limitations of the classical SAR range-Doppler positioning model are more prominent.The observation angle varies greatly,the target scattering presents anisotropy,which makes the scattering information modeling and scattering center extraction are difficult.The development of multi-angle SAR imaging and information extraction has important theoretical significance and engineering application value.In this dissertation,the key issues of spaceborne multi-angle SAR imaging and extraction of geometric and scattering information are studied.This dissertation analyzes the target characteristics of multi-angle SAR images and the key issues in multi-angle SAR processing,revealing that multi-angle SAR has better target information acquisition ability than traditional SAR imaging mode.However,it is still necessary to study how to imaging,and accurately locate and extract target features.Then,this dissertation studies the multi-angle spaceborne SAR imaging.First,by using the high-order range history model,accurate descriptions of spaceborne SAR echo can be obtained with analytical two-dimensional spectrum expression.Based on this,the precise spectrum imaging algorithm with three-step processing is proposed Based on the system parameters and ephemeris parameters of Luojia-02 satellite,an imaging simulation system is established,and the effectiveness of the proposed algorithm is verified by comparing other imaging algorithms.In addition,the problem of SAR geometric location model is studied.SAR geometric positioning accuracy is mainly determined by geometric positioning model and its solution.The classical SAR location model is the Range Doppler(RD)location model.In space-borne SAR data processing,the RD location model must be solved by satellite ephemeris,range function and Doppler function.This requires satellite data products to provide a specific Doppler function,if not,the equivalent doppler function must be calculated to establish the corresponding RD equation,which restricts the application of SAR data products.In this dissertation,the SAR range-history(RH)location model is proposed.It is proved that the RD model is the first-order approximation of RH model.RH model has high accuracy and wide applicability,which can be used as a unified SAR location model.RH positioning model can be applied to SAR data products produced by various imaging processing algorithms without additional Doppler function provided by satellite products.In this dissertation,based on the real SAR data,the accuracy of RH model and RD model is compared.The experimental results show that RH is more accurate in azimuth direction.Finally,the extraction of scattering center for multi-angle SAR targets is studied.The scattering center of SAR image is the core part of the target and the core element of SAR image interpretation.The attribute scattering center model is used to describe the anisotropic characteristics of multi-angle SAR target.On this basis,the problem of scattering center extraction is studied.The scattering center parameter extraction is a process of solving a high-dimensional non-convex optimation problem.The traditional scattering center extraction method is not stable,and the solution depends on image segmentation and parameter initialization.In this dissertation,an extraction method of multi-scattering centers based on real coded genetic algorithm is proposed.The initial parameters can be randomly selected,and the convergence speed is fast.Simulation experiments,electromagnetic scattering calculated data experiments and real data experiments are used to compare the scattering center extraction performance of the proposed method with that of the traditional method,and the advantages of the proposed method are proved.