Adjacent Human Target Accurate Identification And Localization Using Through-the-Wall Radar

Through-the-Wall radar is capable of penetrating obstacles such as glass,wooden boards,concrete,and brick walls,achieving non-destructive detection of hidden targets.It has broad application prospects in urban street warfare,hostage rescue,and postdisaster rescue.However,due to the presence of a large amount of wall clutter,the signalto-noise ratio of the raw reflections of through-the-wall radar is low.In the meantime,the human target belongs to the diffusion target,and imaging of adjacent human targets is prone to be overlapped easily,leading to missing or false detection.How to accurately identify and locate the human target behind the wall has always been a research hot topic in this area.Especially with the expansion of through-the-wall radar technology to the application of multiple adjacent human target identification and localization in complicated detection conditions,current detection technologies are facing various challenges.Currently,a relatively mature set of steps for human target identification and localization using through-the-wall radar has been developed containing three processes,namely,raw reflection acquisition,data pre-processing,and target imaging.To address the issue of precise detection of adjacent human targets in complicated detection conditions,this article conducts profound research from three aspects: improving the quality of raw reflection data received by the through-the-wall radar,suppressing wall clutter under complicated detection conditions,and accurately identifying and locating adjacent human targets.An accurate detection method suitable for multiple adjacent human targets in complex detection conditions is proposed,and the effectiveness and practicality of the new methods are verified through forward simulations and practical experiments.The main research content of the paper is as follows:(1)Aiming to address the issue of low amplitude and signal-to-noise ratio in the reflection of targets received by radar,set increasing gain and decreasing beamwidth of the antenna as the breakthrough point.On one hand,a low sidelobe microstrip antenna is proposed.By applying the slot-loaded technique,the bandwidth of the conventional patch antenna is enhanced.Moreover,the working principle of the E-shaped patch antenna is investigated by analyzing the surface electric field distribution and surface current distribution.Moreover,by conducting the parametric study,the design formulation of the inset-fed E-shaped patch antenna is concluded.On this basis,a low sidelobe series-fed array antenna was developed based on the Chebyshev distribution.The measured maximum realized gain is 25.9 d Bi,and the beamwidths of the E-plane and H-plane are15° and 13.3°,respectively.On the other hand,to further improve the penetration ability of the through-the-wall radar system,an S-band Vivaldi antenna is designed.Surface current distribution is studied to investigate the reason for poor impedance matching performance in the low frequency band of the antipodal Vivaldi antenna.Elliptical loading technology was utilized to improve the impedance matching performance of the antipodal Vivaldi antenna and further reduce the antenna size.The measured maximum realized gain is 9.26 d Bi,and the beamwidths of the E-plane and Hplane are 75° and 46°,respectively.The measurement results prove that both designed antennas have characteristics of high gain and narrow beamwidth,which can effectively improve the received target reflection amplitude and signal-to-noise ratio.In addition,the azimuth resolution of the radar system can be enhanced from the hardware aspect.(2)For through-the-wall radar applications,wall clutter is the most serious interference,which can produce a masking effect on the human target reflection.Moreover,the artifact in traditional BP imaging algorithm seriously affects the accuracy of human target detection.To address the above issues,a through-the-wall radar multiple target detection pre-processing scheme based on WNNM-WCCBP is proposed.First,based on two crucial observations of the reflection of through-the-wall radar,which is the continuity of wall clutter between different traces and the sparsity of the targets in the detection scenario,a through-the-wall radar signal model based on low-rank sparse matrix decomposition is elaborated.Second,a wall clutter suppression method and its derivative methods based on robust principal component analysis are introduced.Third,by analyzing the practical physical significance of singular values of low-rank sparse matrix decomposition in through-the-wall radar detection,a wall clutter suppression method based on WNNM was proposed.In this method,a non-convex weighted kernel norm is used to replace the kernel norm of robust principal component analysis.By specifically shrinking singular values,wall clutter can be more effectively suppressed under complicated detection conditions.Furthermore,to address the issue of strong interference from artifacts of BP imaging,based on the phenomenon that the target reflection has similarity between adjacent traces,a BP imaging algorithm based on weighted cross correlation is proposed,which fully considers the correlation of target reflection between traces.The forward simulation results demonstrate that the proposed wall clutter suppression method enhanced the improvement factor from 35.897 to 63.068 compared to existing methods under complicated detection conditions,and the WCCBP algorithm reduces the integrated sidelobe ratio from-7.543 to-18.941 compared to the traditional BP algorithm,which can effectively reduce the impact of artifacts,improves the focus,resolution,and quality of the target imaging.(3)A K-means clustering and triangulation based method for multiple adjacent human targets detection is proposed to solve the issues of existing technologies being unable to accurately identify and locate adjacent human targets.First,in order to identify human targets using the micro-Doppler characteristic and locate targets through coherent imaging,a 3D human target reflection matrix on the dimensions of range,cross-range,and slow-time was constructed.On the one hand,the human target reflection area can be extracted in the range-slow time plane by using the moving target indicator.The human target’s vital signs and the number of human targets in the detection scenario can be obtained accurately by calculating the variance and mean value of the reflection area.On the other hand,the extracted human target reflections are used for synthetic aperture imaging to acquire the number of human targets in the image.Finally,by comparing the detection results of human targets with these two methods,human target localization methods based on K-means clustering and triangulation are proposed,respectively.In addition,even though only 59% of the traces in a typical detection scenario can accurately identify human targets,an accurate number of human targets can still be obtained,proving that the proposed method has a high human target detection rate and robustness.The forward simulation results indicate that the proposed method has a localization error of less than 12 cm under complicated detection conditions where the spacing between adjacent human targets is 30 cm and there is an angle between the antenna and the nonuniform wall.Finally,two different types of through-the-wall radar systems were constructed using the designed antennas,and real-world through-the-wall radar multiple human target detection experiments were conducted in urban buildings using the proposed method.The experiments comprehensively considered the detection conditions including human targets with short spacing and different orientations,complicated wall structures,and the antenna and the wall are unparalleled.The results indicate that the radar system constructed in this article and the proposed multiple human target detection method can effectively and accurately identify human targets and accurately locate them.The measured localization errors of these two through-the-wall radars are less than 10 cm and17 cm.respectively.In addition,the proposed WNNM-WCCBP multiple human target detection method can effectively suppress complex wall clutter,improve imaging quality decently,and weaken the influence of artifacts.It fills the research gap of existing technologies in such issues and provides effective support for accurate human target detection in complicated detection conditions.