Research On Application Of Millimeter Wave Radar Imaging Based On 77GHz

Because millimeter wave has the advantages of being able to achieve penetrating detection and all-weather work,millimeter wave imaging has become an irreplaceable solution in related imaging fields.In security imaging scenarios,millimeter waves can penetrate clothing to image potentially dangerous goods without generating ionizing radiation;in assisted driving scenarios,millimeter waves can adapt to inclement weather and are not affected by light intensity,providing a guarantee for vehicle driving safety.With the improvement of millimeter wave radar design level and production technology,the accuracy and integration of radar systems are getting higher and higher,which provides convenience for the research and application of millimeter wave imaging technology.This paper uses the77 GHz frequency modulated continuous wave radar as a platform to carry out the application research of millimeter wave imaging in the above two scenarios.The specific work content of this paper is as follows:(1)The influence of the spatial position of the millimeter wave radar array element and the waveform parameters on the reconstructed image is studied.A hardware platform is built based on the 77 GHz AWR1843 radar and two-dimensional rails,and MATLAB software is used to develop a system program that can be used for automatic data acquisition.The system provides experimental conditions for the subsequent data acquisition of security imaging scenarios and automotive millimeter wave imaging scenarios.(2)The use of sparse arrays in the security scenarios can effectively reduce the system cost,but the reconstruction method based on compressed sensing takes a lot of time and hinders the real-time performance of imaging.This paper studies it from the perspective of signal recovery and designs a sparse imaging scheme based on deep learning method.Existing single-channel or three-channel network models cannot directly train and predict complex signals.To solve this problem,a Unet network based on complex convolution is constructed to realize the recovery from under-sampled complex signals to full-sampled complex signals,and the target is reconstructed using a two-dimensional matched filter algorithm of distance-dimensional pulse compression.Reconstruction experiments on under-sampled simulation data under different sparsity and sampling intervals show that the scheme can effectively get the target image.Using measured data for verification,in different sparsity,based on the deep learning method can complete the imaging in a very short time,and has a relatively satisfactory effect.(3)In the automotive millimeter wave radar imaging scenarios,in view of the problem that the existing one-dimensional automotive synthetic aperture radar imaging schemes can not distinguish the elevation information of the target,a two-dimensional array of automotive millimeter-wave radar imaging scheme is designed.The direction uses the vehicle movement to form a synthetic aperture,and the elevation direction uses multiple input multiple output methods to obtain an equivalent uniform array.Experimental results prove that the imaging results of this scheme can effectively reflect the profile of the target.In the scene of the non-linear trajectory of the vehicle,there is the problem that the time complexity of the only available backward projection algorithm is too high.According to the characteristics of the array elements of the above imaging scheme,a fast imaging algorithm combining time domain and wave number domain is proposed.Theoretical derivation and simulation experiments show that the algorithm can effectively reduce the time complexity and maintain a good imaging effect.