Millimeter-wave Passageway Multi-input Multi-output Imaging System Development And Algorithm Research

Scientists and engineers have been working hard in exploring frequency bands of electromagnetic wave,including radiation sources,detectors,detection applications and algorithms for a long time,and great achievements have been made.To date,a variety of security screening approaches have been developed almost over all frequency bands that can be achieved by hardware technology.However,X-ray imager remains the best behaved in terms of detection ability: Strong material-penetration capability with higher SNR echo makes it easy to achieve tomography;Short wavelength gives rise to better spatial resolution and higher levels of radiation energy,so X-ray damage to living cells is often criticized in security imaging.Therefore,the detecting energy with strong penetration and high resolution but with low radiation dose is demanded.Luckily,millimeter wave can meet these requirements to some extent.Millimeter wave locates in 30-300 GHz frequency band,i.e.,1-10 mm wavelength,right between microwave and terahertz wave.Millimeter wave is able to penetrate common clothing materials(especially in the low frequency band of millimeter wave),but reflects strongly by living skins achieving body surface imaging;Common metal and various non-metal(explosives,plastics,ceramics,etc.)materials can be depicted effectively with the spatial resolution of centimeters or millimeters;Far less radiation dose than that of mobile phones,it makes no harm to human body.The energy of millimeter wave impinging on living body surfaces is reflected and received by the antennas.The target image is reconstructed by some computational approaches and the features of the target are identified by image processing methods.A passageway MIMO millimeter-wave imaging system is proposed in the dissertation.In the light of system development and algorithm design,the main research content of this dissertation is as follows:Firstly,the mathematical model for radar imaging is derived from the electromagnetic scattering problem.Based on this model,the three-dimensional pulse compression method is formulated in the light of Wiener filter.Nyquist’s sampling law and resolution estimation are given quantitatively for MIMO array.In terms of spatial sampling,the frequency diversity MIMO array is proposed.For the millimeter-wave passageway imaging system,the doppler effect caused by human body movement in the near field is too complicated to compensate,and real-time data acquisition is more practical to suppress doppler frequency shift.Since time division is not applicable for large phase error(long sampling time,at least one frequency-sweep cycle or pulse period multiplied by the number of transmission channels),frequency diversity is recommended for the prototype;With frequency diversity combined with single snapshot imaging,in a particular period,all the array elements for a single snapshot transmit detection waves with different frequencies.Ideally,a single snapshot sampling can be completed in one sweep period,significantly reducing the sampling time.In MIMO system array design and optimization,we propose a scanning passageway constructed from MIMO array periodic extending.Generally,the elements of the expanded array retain the property of uniform sampling.Therefore,RMA can be used in principle to accelerate image reconstruction for real-time performance.The characteristics and optimization strategies of sparse arrays are studied preliminarily.Sparse MIMO arrays are designed to minimize the number of array elements(and channels)with confined image quality degradation.The optimization strategy adopted in this paper is: In the light of helical arrays,optimize the array performance on a factor,which controls the geometry of the spiral MIMO array,through optimization(local or global optimization).In particular,for a periodic passageway imaging system,we develop fast and high precision imaging methods for the proposed periodic passageway imaging system.Firstly,regarding the proposed passageway imaging system composed of dual MIMO arrays,RMA is extended to the dual MIMO array imaging scheme to ensure the real-time performance of the 3D image reconstruction algorithm.The innovation of this method is divided into four cases according to the transmitting-receiving mode,among which the main difference is the wavenumber of down range.Thus,the center scene compensation and Stolt interpolation should be performed respectively.The final image is the summation of the products of the four modes.Furthermore,although MIMO RMA can be used for sparse arrays with regular distribution in principle,the inconsistency of sampling points in the down-range dimension increases the interpolation error,resulting in a reduction in the dynamic range of the final image and an increase in the background noise.In this paper,a dimensional-factorized RMA(DF-RMA)is proposed.The scientific contribution is to decompose the original formula into multiple RMA programs on sparse dimensions.The reflectivity results are supposed to account for multiple cross array combinations,and the final image can be obtained by coherence stack of reflectivity results.As demonstrated in the simulation results,DF-RMA improves the dynamic range of the image and the anti-noise performance of the system significantly with limited computation increase.To explore the capability of the system,super-resolution imaging methods are studied in this dissertation."Super-resolution" imaging methods aim to improve the imaging quality from image reconstruction algorithms,including spatial resolution,side-lobe level and dynamic range.This paper begins with MIMO PSM--phase shift migration algorithm(PSM)in MIMO arrays.As only FFT and phase compensation are included in the algorithm,so it is easy to obtain the adjoint operator.Combining with total variation,a high-performance regularized method can be derived,which is considered as a classical iterative super-resolution imaging method.For the proposed system,the performance of real time and image quality acts equally important.In this paper,a coherence factor-based RMA(CF-RMA)is proposed,which significantly improves the dynamic range and spatial resolution of images,and the computational load is only equivalent to that of two independent RMA programs.The contribution is that the double integral of the reflectivity noncoherent power is approximated by a single integral in the light of point matching method.Moreover,a data-rearrangement approach of low computational cost in wavenumber difference matrix is proposed to derive a new integral kernel,through which the formulation of the reflectivity noncoherent power calculation can be implemented in the light of RMA program.Finally,a cross MIMO array imaging testbed based on a vector network is designed and developed.Firstly,the imaging performance of the array geometry is verified through electromagnetic simulation in terms of imaging resolution and quality,including images of handgun alone,human body alone and the handgun carried by human body,which proves the capability in security imaging.The experimental testbed developed in this dissertation is used to carry out 3-D imaging of point target and several metal targets,and the performance comparison of MIMO RMA and MIMO CF-RMA is demonstrated with the imaging results,validating the 3-D imaging capability of MIMO array-based millimeter-wave imaging system.