Ultra Wideband Radar Image Reconstruction Technique Based On Compressed Sensing

Recently,the ultra-wide band(UWB)radar has attracted more and more attentions owing to its extremely wide signal transmission band and great application potential in target detection,imaging and recogtion.With its further development towards multi-channel,multi-polarization,multi-band and high-resolution,UWB radar system is facing several challenging problems,such as the huge amount of measurement data and long data acquisition time,which severely limits the widespread application of UWB radar system.In this thesis,the UWB radar imaging method based on compressed sensing(CS)theory is employed to solve the existing problems of UWB ground penetrating radar(GPR)and through-the-wall radar(TWR)system in the fields of system construction and imaging reconstruction.The specific contents and main achievements of this thesis are presented as follows:(1)In order to solve the invalidation of the traditional CS SFCW-GPR migration imaging method in strong clutter environment,a CS-based SFCW-GPR migration imaging method based on the subspace projection clutter suppression technique is proposed.Firstly,the original uniform sampling data in the frequency domain is reconstructed by the compressed measurement model at each antenna measurement position.Then,the subspace projection clutter suppression technique is employed to filter out the strong ground echo.Finally,the sparse reconstruction algorithm is used for CS imaging reconstruction of the underground target image.The imaging results of simulation and experimental data show that the proposed image reconstruction algorithm can realize accurate imaging localization of the underground target in the strong clutter environment with only 9.8%reduced data.(2)The existing CS based GPR migration imaging methods are usually computationally intensive and sensitive to the regulariton parameters.To solve the problem,an imaging method based on the Bayesian compressive sensing framework is proposed with SFCW-GPR as an example.The reflection coefficient of observed scene is reconstructed by establishing a hierarchical sparse model and a relevance vector machine.The simulation results show that,compared with the traditional CS imaging method,the proposed imaging method can utilize the statistical prior information of the observed scene to balance the reconstruction precision and calculation efficiency more effectively.Furthermore,the GPR migration imaging algorithm based on multi-task Bayesian compressive sensing is studied to solve the problem of low signal-to-noise ratio and low data volume.The simulation results exhibit that,under the condition of low signal-to-noise ratio and low data volume,the reconstruction accuracy of multi-task Bayesian compressive sensing is higher than that of the single-task Bayesian compressive sensing.Multi-task Bayesian compressive sensing further improves the accuracy of imaging results.(3)In oder to overcome the drawbacks of high bandwidth,large aperture and long measurement time of the high resolution GPR Diffraction Tomography(DT)imaging method,the multibistatic DT imaging reconstruction method based on CS framework is proposed.This new method uses compressive sampling to collect the frequency domain scattered field data of underground target.Then,according to the sparseness of received signal at each antenna measurement position,the original time domain echo signal is reconstructed with the sparse reconstruction algorithm and the full set of scattered field data in the working band is recovered.Finally,the traditional multibistatic DT algorithm is used for imaging reconstruction.Furthermore,the Bayesian Compression Sensing(BCS)algorithm based multiview-multistatic DT imaging method is proposed.The correlation of frequency domain measurement data is applied to reconstruct the missing frequency data in all measurement data,which improves the data acquisition speed.The imaging results of the full-wave electromagnetic simulation data generated by finite-difference time-domain method show that the proposed imaging algorithm can ensure the resolution and accuracy of the target image even in the case of 20%and 12.9%compression ratio for multibistatic and multiview-multibistatic configuration,respectively.So the on-site measurement speed of the system can be greatly increased by the proposed imaging scheme.(4)A CS based TWR imaging method in multi-path environment is proposed.When the paremeters of wall is unknown,a imaging method of expressing the wall parameters estimation and the target reconstruction as a joint optimization problem is proposed.This study takes the UWB impulse TWR system as an example.Firstly,the multi-path echo signal model of the TWR system is established by the electromagnetic wave specular reflection principle in the observed scene with known wall parameters.Based on the analysis of target echo,a block sparse imaging model of the observed scene is established.Then,the imaging algorithm of reconstructing the stationary and moving targets are presented in the multi-path effect environment.The simulation results show that the proposed imaging method enables accurate high-resolution imaging of the concealed target in the multi-path detection environment.In the case of wall parameters being unknown,estimation of wall parameters and imaging reconstruction are thus expressed as a joint optimization issue.The nonlinear optimization algorithm and the block sparse reconstruction algorithm are used for solving the joint optimization issue,estimating the wall parameters and achieving the imaging reconstruction.The simulation results show that not only accurate estimation of the wall parameter is achieved,but also the high resolution and high accuracy of the target imaging reconstruction are realized.This study greatly extends the applicable scope of the TWR CS imaging method.For various applications of CS-based GPR and TWR imaging,this thesis focuses on how to use CS for GPR migration imaging in the strong clutter environment.This thesis solves the problems of the high computational complexity in traditional CS GPR migration imaging method and proposes the CS based GPR DT imaging framework.A CS-based TWR imaging method with unknown wall parameters in multi-path environment is proposed.The results of this research are of great significance for the application of CS-based UWB radar imaging engineering.