Research On Microwave Computational Imaging Technique Based On Metamaterial Aperture

Microwave Computational Imaging based Information Metamaterial Aperture(IMA-MCI)is an emerging radar imaging technology.Compared to relatively mature Synthetic Aperture Radar(SAR)and Inverse Synthetic Aperture Radar(ISAR)technologies,IMA-MCI does not rely on the relative motion between the radar platform and the target,and can achieve high-resolution imaging within the beam.It breaks through the limitations of traditional radar imaging technologies that require large real apertures or synthetic apertures.IMA-MCI has the ability to perform forward-looking imaging,which is mainly dependent on the excellent modulation ability of information metamaterial antennas for electromagnetic waves.During imaging,this antenna can form a space-irrelevant radiation field in the target imaging region.By repeatedly illuminating the target imaging scene,IMA-MCI can reconstruct the target using computational imaging methods.Therefore,the coding measurement mode and computational imaging algorithm play a crucial role in imaging performance.This article aims to improve the imaging performance of IMA-MCI by constructing a computational imaging method based on deep learning and phase retrieval,and designing a coding optimization scheme that combines the hardware of coding aperture with the backend computational imaging method.This provides a theoretical basis for the practical engineering applications of IMA-MCI technology.Firstly,the flexible regulation ability of information metamaterial antenna on electromagnetic wave is verified by using electromagnetic simulation software ANSYS HFSS,the imaging model of IMA-MCI is established,and the target reconstruction is completed by using classical imaging algorithm.It includes orthogonal matching tracking algorithm(OMP),total variational augmented Lagrange alternating direction algorithm(TVAL3)and sparse Bayesian learning algorithm(SBL).Secondly,the reconstruction of the target scene by IMA-MCI is still challenging in the presence of phase error and low Signal to Noise Ratio(SNR).To address these issues,a microwave computational imaging model based on reflective information metamaterial antennas was established,and a computational imaging method based on deep learning and phase retrieval was proposed.Simulation results showed that this method has good scene robustness.Finally,in order to further improve the resolution of the imaging,an end-to-end neural network was constructed based on the imaging model of the transparent metamaterial antenna,which can couple the hardware coding aperture mode design and the backend computational imaging method into a single network for joint optimization of the two parts.The effective combination of the hardware measurement stage and the backend computational algorithm realizes the efficient and accurate reconstruction of different sparseness targets under compressive measurement.Simulation experiments show that this method can reconstruct targets with different sparsity degrees from compressed measurements,and has good anti-noise performance,thereby improving the recovery ratio of compressible targets.