Studies On One-Bit Coded SAR Sparse Imaging With Time-varying Thresholds

Synthetic Aperture Radar(SAR)is a kind of high resolution imaging radar,which has been widely used in military and civil fields.Due to the limitation of Nyquist sampling theorem,a large amount of radar echo data is received for high resolution SAR images,which will bring serious burden to the data storage and transmission at the receiver,causing radar system to be very complex.Different from traditional signal sampling methods,data compression and sampling are achieved simultaneously in the compressed sensing framework,and the amount of data will be reduced.The recent theory of compressed sensing proves that the sparse signals can be accurately reconstructed with much less measurements than Nyquist sample rate's need.However,conventional CS framework still needs high speed ADC to acquire high precision echo data,which will bring about greater cost of hardware and energy consumption to radar system.One-bit compressed sensing proves the direction of sparse signals can be reconstructed with sign information of signals.A significant advantage of one-bit quantization is that it can be realized by a simple comparator which is faster,cheaper and lower energy consumption than ADC.One-bit compressed sensing with zero thresholds will lose the amplitude information about signals.Therefore,Time-varying thresholds have received attentions recently,which is chosen in one-bit quantization for retaining the amplitude information of the reflectivity coefficient of the targets.One-bit compressed sensing SAR imaging with time-varying threshold is studied in this paper.Some problems that the method may face are discussed,such as how to choose the appropriate quantization threshold,and when the imaging scene is large,the algorithm complexity is too high.For the above two problems,the solution given in this paper is as follows:Our first content in this work is one-bit SAR imaging algorithm with time-varying thresholds based on logistic regression.The one-bit quantization is considered as a linear classification procedure,and the logistic regression algorithm with Li norm regularization is used to recover the original reflectivity coefficient of the sparse targets in scenes.Simulation results verify the effectiveness of the algorithm.On this basis,a new adaptive threshold selection method based on optimization algorithm is proposed.The simulation results show that one-bit SAR imaging method with the adaptive threshold has more accurate and stable imaging results.The second content in this work is one-bit SAR imaging algorithm with time-varying thresholds based on approximate observation.The large matrix storage and its multiplication are replaced by the matched filter operator and the inverse matched filter operator,which effectively reduces the time and space complexity of the algorithm.The simulation results verify the effectiveness of the proposed algorithm.A feasible solution about one-bit SAR imaging algorithm with time-varying thresholds applied to the large scenes imaging is provided.