Technique Of Human Target Localization And Characteristic Extraction With Ultra-wideband Through-the-wall Radar

By transmitting a specific low frequency of electromagnetic waves,to penetrate wall and other non-metallic obstacles,and receives echo of hidden targets,Through-the-wall Radar(TWR)is widely used in military and civil applications,especially in urban battle,anti-terrorism,and disaster rescue.The low reflectivity of human body,strong environmental clutters,and high signal attenuation of walls result in fairly low signal-to-clutter-and-noise ratio(SCNR),especially for micro-motion human target with the low-frequency and low-amplitude motion.In this paper,we focus on the location and feature extraction of the micro-motion human target behind wall,which covers several key technical problems of the micro-target location using ultra-wideband(UWB)TWR,including micro-motion target location based on channel detection,micro-motion target location based on imaging,micro-motion target location under multipath environment,and human life signal extraction.The main work and contributions of this dissertation are as follows:Firstly,based on the analysis of human vital sign,UWB signal,and electromagnetic wave propagation model in TWR,a space-time-frequency joint processing framework is established,in which the space domain is the display of the antennas,the time domain is the frame serial,and frequency domain is the signal transmiiting frequency.By making full use of space-time-frequency information,we achieve the detection,imaging and location of of micro-motion targets behind wall.Secondly,with only a limited number of data channels of TWR,we proposed a micro-motion target location method based on target detection in each channel.The micro-motion target is firstly detected in each channel,then we combine the detection results of all channels to locate target.In the detection processing,a low rank and sparse matrix decomposition(SLRMD)based method is proposed to suppress non-stationary clutter in range-time domain,simulation and experimental results validate its effectiveness to suppress non-stationary clutter.When there are multiple stationary human targets,wrong data association may result in localization of false alarms.We propose an approach of multiple stationary targets association method based on data correlation using one transmitter and two receivers.Then target localization and wall parameters estimation are transformed into a solution of constrained optimization problem when considering the measured noise.The localization performance of the proposed method is evaluated with different wall parameters.Then,we propose a micro-motion target localization method based on imaging processing when using TWR imaging radar.In space-time-range domains,imaging of every frame is firstly performed,based on the fact that the fourth order cumulant(FOC)of noise of Gaussian distribution are zeors and the linearity of back projection imaging,the FOC of complex noise of Gaussian distribution is also zeros,then a FOC processing method is proposed to imaging and locating micro-motion targets,the proposed method can improve the SNR remarkably compared with traditional methods.Then in space-Doppler-range domains,we propose a Doppler slice based method to image and locate micro-motion tareget.For mciro-motion targets,the imaging slices of the Doppler frequency of interest can separate micro-motion targets with different vibrating frequencies.Simulation and experimental results validate the proposed methods.At the same time,aiming at the detection and location of micro-moving targets in indoor multipath environment,the multipath propagation model of electromagnetic wave is established,including the wall ringing multipath inside the wall and the multipath betewwn the interior walls,and an association between multipath ghost and real target are established under the condition of the known structure of the building.According to this association,a multipath exploitation method is proposed to enhance the SCNR of the real target under multipath einvronment.The method is verified by theoretical simulation,FDTD simulation and measured data.Finally,to extract the vital sign from human micro-motion,as the micro-motion includes human vital sign as well as random body movement.An isophase based method is proposed to exact human micro-motion locus.We use the extend EMD(EEMD)based method to decompose micro-motion into the trend movement and the periodic movement,and the fromer corresponds to human body random movement while the latter corresponds the human vital sign(respirationary movement).In order to extract vital sign under low SCNR conditions,a multi-channel singular spectrum analysis(MSSA)based method is proposed to decompose micro-motion using echoes from multiple range cells.The echoes are also decomposed into the trend movement and the periodic movement.Experimental results with real data show that human's respiration movement can be accurately extracted under the condition of random body movement.Finally,we give summarizes of the whole paper,and point out the direction of further research in human target detection and feature extraction.