Research On Technology Of Three-dimensional Terahertz Coded-aperture Imaging

With the advantages of both high resolution for optical imaging and strong penetrability for microwave imaging,terahertz radar imaging has gradually become one of the research focuses of high-resolution radar imaging.However,the principle of current terahertz radar imaging relies on the relative motion between the radar and the target,which requires much scanning time or high system cost.To overcome the disadvantage of conventional terahertz imaging,this paper proposes an imaging method named terahertz coded-aperture imaging(TCAI),which is based on the basic principle of terahertz imaging,microwave coincidence imaging and computational imaging.TCAI uses the coded metamaterial to modulate the terahertz beam,and thus the target scattering information in the echo signal is coded and indexed.Finally,the target reconstruction methods,such as compressed sensing,can help TCAI achieve high-resolution radar imaging under the high-resolution,forward-looking,and staring imaging.Therefore,TCAI and conventional terahertz imaging can replenish each other in high-resolution imaging.TCAI holds the advatantages of high resolution,high frame frequency and low cost,which can be widely applied into security check,antiterrorism,battlefield reconnaissance and medical imaging,etc.This paper studies mainly on the technology of three-dimensional(3D)-TCAI(3D-TCAI),which includes the imaging principle,coding optimization and 3D imaging methods.Chapter 1 expounds the research background and significance of this paper,reviews the research status of terahertz radar imaging,coded-aperture antenna and coded-aperture imaging technology,and thus points out the necessity of research on 3D-TCAI.Chapter 2 studies the imaging principle,mathematical model,and coding-optimization scheme of TCAI.Firstly,this chapter introduces the basic principle and imaging model,expounds the high-resolution reason of coded-aperture imaging,and analyzes the imaging performance of various reconstruction algorithms based on sparse and expanded targets.Secondly,the coding strategies are categorized according to different coding positions,objects and patterns.Referring to different coding strategies,this chapter builds the coded-aperture imaging models,compares and analyzes their high-resolution imaging ability.Besides,the main influencing factors of coded-aperture imaging factors are also analyzed and summarized.All works in this chapter lay the groundwork for 3D-TCAI.Chapter 3 studies the imaging method of 3D-TCAI based on spatial-grid partitioning,which derives from planar-grid partitioning.Firstly,We present both the basic principle of two-dimensional(2D)and 3D imaging methods based on planar-grid and spatial-grid partitioning,respectively.The simulation results of simple 3D sparse targets validate the imaging feasibility of 3D-TCAI based on spatial-grid partitioning.To achieve high-resolution imaging for electrically large and complex targets in near field,we propose the basic design method of quasi-optical scanning for 2D imaging,and further propose the optimization method of quasi-optical scanning for 3D imaging,which can scan the large-scale imaging area block by block,decrease the grid-cell number of each imaging.Thus,the computational complexity is efficiently reduced.Finally,based on quasi-optical block scanning,the 3D imaging method of multiple resolutions is proposed to achieve Near-field and high-resolution 3D imaging for human object.Chapter 4 studies the imaging method of 3D-TCAI based on range-domain(RD)slice,which is obtained by the principle of pulse compression.To solve the problem of high computational complexity for 3D-TCAI.This imaging method firstly divide and extract the back signals of different RD slices and reconstruct the 2D imaging planes in parallel,which avoid the grid-cell division in range dimension,and thus reduce the computational complexity greatly.Firstly,this chapter introduces the basic principle and imaging model of RD slice based 3D-TCAI.To obtain the RD slice from different transmitting signals,this chapter presents how to process the linear frequency modulation(LFM)and hopping frequency(HP)signals by dechirping and matched filtering,respectively.The simulation verifies that the RD slice based 3D-TCAI can achieve high-efficiency,high-accuracy target reconstruction under low signal-to-noise ratio(SNR).However,there are still some false scattering information under extremely low SNR.Therefore,we propose an enhanced imaging method based on convolutional neural network(CNN),based on which the reconstruction accuracy is further improved.As the RD slice is hard to be recognized under terrible SNR,this chapter proposes the auxiliary imaging method based on information geometry.The simulation results prove that the RD slices under low SNR can be extracted by the information geometry based imaging method,which improve the RD slice based 3D-TCAI.Chapter 5 studies the imaging method of 3D-TCAI based on simultaneous reconstruction by both RD slice and plane division,the obtaining method of RD slice is similar to chapter 4.Although the RD slice based 3D-TCAI can reduce the computational complexity in a certain degree,the electrically-large 2D imaging plane and high-resolution grid cells results in heavy computational burden.Firstly,this chapter introduces the imaging principle,procedure and mathematical model of 3D-TCAI based on simultaneous reconstruction by both RD slice and plane division,which can independently extract the space-domain(SD)echo signals corresponding to different subareas of the same imaging plane,and effectively solve the problem of large-sized reference signal matrix(RSM).According to the dividing approaches of SD echo signal,this chapter further proposes three kinds of imaging methods based on correlation operation,back projection(BP)and Fourier transformation(FT),separately.Compared with the imaging methods of 3D-TCAI based on spatial-grid partitioning and RD slice,the simulation results demonstrate that the imaging in this chapter can achieve higher reconstruction accuracy with much lower computational complexity.Chapter 6 summarize the main research achievements in this paper,and plans the future work.