Research On Approximated Observation-Based Compressive Sensing Radar Imaging

In the field of radar imaging,the application of compressed sensing(CS)can reduce the sampling rate of radar system,reduce the volume of echo data,and improve the image quality dramatically.Although the application of CS will bring a lot of advantages,there are some disadvantages,such as high computational complexity and much memory cost.Then,the approximated observation-based CS radar imaging method emerged as required.This method not only can contain the inherent advantages of CS,but also will reduce the computational complexity and the memory cost dramatically.Therefore,there will be important theoretical value and application potential to explore the problems in practical application about the approximated observation-based CS radar imaging.In this dissertation,based on the research of the approximated observation-based CS radar imaging method,we have explored the problems in practical application,for example,the radar position uncertainties will bring the phase error,and the CS radar imaging method cannot be applied directly when the imaging scene is not directly sparse.Then,this dissertation is organized as follows.In Chapter one,the history of development about synthetic aperture radar(SAR)is first introduced.Then,the theory of CS and its advantages are introduced.Finally,we introduce the significance of studying the approximated observation-based CS-SAR imaging and the challenges which will be encountered in the applications of this method.In Chapter two,the elementary knowledge,the properties of sensing matrix and the reconstruction algorithms in the theory of CS are first introduced.Next,we compare the advantages and disadvantages of the approximated observation-based CS-SAR imaging method with the conventional CS-SAR imaging method and the matched filter-based method.The comparisons are focused on four aspects:memory cost,computational complexity,imaging quality and whether it is necessary to satisfy the CS conditions.Then,aiming at the shortcomings from existing methods of stepped-frequency SAR imaging,the approximated observation-based CS-SAR imaging method using stepped-frequency waveform has been proposed.The proposed method not only will contain the inherent advantages of CS-based method,but also can reduce the computational complexity and the memory cost compared with the conventional CS-SAR imaging method.Finally,the simulation results validate the effectiveness of the proposed method.In Chapter three,the signal models of SAR imaging and the phase error models arising from the radar position uncertainties are first introduced.The signal models contain linear frequency modulated(LFM)waveform,stepped-frequency waveform and random-frequency waveform.Then,the problems of phase error and observation model mismatch arising from radar position uncertainties have been solved.There are some disadvantages about the existing phase error correction methods for CS-SAR imaging,such as high computational complexity and much memory cost.In order to deal with these shortcomings,the phase error correction method for approximated observation-based CS-SAR imaging is proposed.Compared with the existing phase error correction methods,the proposed method can reduce the computational complexity and the memory cost significantly.Finally,Simulation and experimental results validate the effectiveness of the proposed method in the presence of either one-dimensional(1D)phase error or weighted 1D phase error.In Chapter four,both the signal model and the approximated observation model of SAR imaging are first briefly reviewed.Then,we investigate the situation in the presence of both sparse targets and extended targets,implying that the theory of CS cannot be applied directly,and that the imaging scene will need to be sparsely represented.Although some published literatures have studied the mixed sparse representation(MSR),there are some disadvantages about the existing MSR method.Aiming at these disadvantages of the existing method,a new MSR method has been proposed.Finally,the simulation and experimental results show that the proposed method not only can deal with the comp lex-valued SAR images effectively,but also will demonstrate superior performance over the Chirp Scaling algorithm and the existing MSR method.In Chapter five,the situation in the presence of both phase error and non-sparse scenes has been dealt with.There have been some existing imaging methods which can achieve phase error correction for CS-SAR imaging,and there also have been some published literatures which can deal with the situation in the presence of both sparse targets and extended targets.However.when both two problems are encountered at the same time,there has been no way yet to solve them jointly.Therefore,phase error correction method for MSR-based CS-SAR imaging has been proposed,which not only can deal with the situation in the presence of both sparse targets and expended targets in the radar imaging scene,but also will correct phase error arising from radar position uncertainties effectively.Finally,simulation results validate the effectiveness of the proposed method.The Chapter six summarizes the whole dissertation and looks forward to the future work.