Research On High Resolution SAR Imaging Technology In Wavenumber Domain

Synthetic aperture radar(SAR)is a kind of coherent radar for high resolution imaging.Since it receives the echoes of the transmitting waves without considering the influence of visible lights,SAR is able to work under all weather conditions,day and night.It has been widely used in many aspects such as military reconnaissance,topography surveying,environmental monitoring and disaster evaluation.Due to the requirements of different applications,various SAR signal processing methods have been developed for SAR imaging algorithms.As one of them,the sub-aperture imaging algorithms are promoted in the real-time imaging systems.Because of the characteristics of short accumulative time and lower computation,the sub-aperture imaging has advantages for imaging complexity,efficiency and motion compensation(MOCO).So the quick look processing can be realized by applying the sub-aperture imaging to the parallel digital signal processing systems with high performance.Meanwhile,in order to make detailed observations of the interested scenes,the high resolution images can be achieved by coherent additions of the sub-images.During the processing above,the approximation of the imaging algorithm,the coherence of the sub-data and the accuracy of the MOCO together make a directly influence on the imaging quality and performance of image fusion.So this dissertation concentrates on the realization of the real time imaging system with high resolution.And researches have been done on the sub-aperture imaging algorithms,the fusion of the sub-data,the proper accuracy MOCO approaches and the fast raw data simulation.The main content is summarized as follows.1.For the requirement of the real time SAR imaging system with high resolution,a modified Omega-K algorithm for squint SAR sub-aperture imaging is developed,which is based on the azimuth resampling technique.The Omega-K based algorithms can sampling the 2-D wavenumber spectrum(WS)uniformly by applying the Stolt or extended Stolt interpolations.Therefore,the range cell migration(RCM)can be corrected without approximation.It is completely accurate in theory.For the requirement of the real time imaging,the time-consuming interpolation processing can be replaced by the non-uniform fast Fourier transforms(NUFFT).In order to make the algorithm adapted to the high squint sub-aperture imaging,by analyzing the skew WS,the azimuth resampling technique and the coordinate rotation approach are implemented to realize the “no squint” procedure.Such process simplifies the traditional WS expression and extends the available data support region(DSR),improving the imaging quality.For the problem of the short DSR of the sub-aperture data in azimuth distance domain,the proposed algorithm apply the extended Stolt interpolation to solve the range-azimuth coupling,meanwhile retain the azimuth modulation term,finally focusing the azimuth data in wavenumber domain.By this mean,the azimuth processing for sub-aperture data avoids the large number of zeroes padding in traditional Omega-K algorithm.2.Based on the sub-aperture modified Omega-K imaging algorithm,the sub-image mosaicking and fusion processing is proposed by analyzing the coherent relation between sub-aperture data.To realize the high speed parallel processing,the sub-aperture imaging operation should be independent with each other,and there exists differences between each sub-aperture including the targets coordinate difference and coherent phase difference.Before the coherent addition procedure,the positon compensation functions are constructed to unify the coordinates in different sub-aperture of the same targets.For the problem of phase difference,the globalization processing is implemented after the compensation of constant phase terms.This procedure not only eliminates the aperture dependent phase but also places the sub-aperture distance DSRs to their corresponding section.After the coherent addition procedure,the distance DSR of sub-data would be extended,improving the azimuth resolution.Since it can select the sub-data freely to implement the coherent addition operation,the proposed algorithm is much more flexible.3.Because of the sensibility of the high resolution imaging system to the motion error,an improved angle correction SATA and a MET-WD millimeter waves(MMW)MOCO approaches are developed for the along-track and cross-track motion error,respectively.Due to the WS deviation caused by the along-track motion error,the ideal time-frequency mapping relation is less accurate,thus,is insufficient for the MMW SAR system imaging.By introducing an instantaneous squint angle correction,the accuracy of the time-frequency mapping relation is increased.Then,combined with the improved mapping relation,the modified SATA is suitable for the aperture-dependent along-track MOCO.For the cross-track motion error,in order to solve the problems of the “ghost” target,resolution loss and non-systematic range cell migration(Ns RCM)which are introduced by the residual error after the first step compensation in TSA,a MMW MOCO approach based on MET is proposed.By applying the error equivalent mapping technique,the MET compensate the range-independent,range-dependent and part of the aperture-dependent phase errors,eliminating the influence of the residual error on the 2-D WS.To further satisfy the requirement of the MMW system imaging,after the RCM correction(RCMC),the wavenumber division(WD)method is done to compensate residual aperture-dependent motion error.4.On the basis of the extended Omega-K algorithm,a raw data simulation based on the inverse extended Stolt interpolation method is developed.Considering the Stolt based interpolation could realize the RCM correction(RCMC)without approximation,based on its inverse process,an RCM recovery approach is proposed.At the same time,introduction of the motion error into the separated azimuth phase term could be implemented.For the aperture-dependent motion error,by dividing the azimuth wavenumber into blocks and introducing the phase error for each of them,the completed aperture-dependent motion error can be achieved after the addition of all blocks.