Research On Human Identification Theory And Technology Based On Non-contact Vital Sign Detection Radar

Non-contact vital sign detection radar can identify the position,physiological state and other features of human subjects by extracting their vital signs and life activity information.Compared with traditional video and audio sensors,non-contact vital signs detection radars have advantages of great environmental adaptability,ability to detect through obstacles and less privacy issues.Thus,it has become a promising research topic in both modern radar and biomedical sensing fields.It has great potential for health care,anti-terrorism and rescue applications.However,there are still some issues to be solved in the field of human identification based on non-contact vital signs detection radar.On this basis,this dissertation focuses on multi-target through-wall detection,two-dimensional(2-D)imaging,human target identification and the in-depth study of human identification theory based on non-contact vital sign detection radar.The contributions of this dissertation and innovations are shown below:(1)To solve the problem of mutual interference and signal aliasing for throughwall human target identification.This dissertation studies the through-wall multi-target vital sign separation technology and proposes a through-wall time-varying vital signs separation algorithm based on variational mode decomposition(VMD).The algorithm fully considers the position and breathing condition of human subjects behind the wall.Through the modal decomposition and time-frequency analysis of the echo signal,the method can not only locate multi-target though-wall accurately,but also distinguish human targets within the same range-bin.The experimental results show that compared with the traditional vital sign detection method,the proposed algorithm can effectively separate the vital sign signals of three targets through-wall,showing great performance and applicability when dealing with the signal aliasing and interference problems in through-wall multi-target detection.(2)To solve the human target identification problem under the mobile radar platforms.This dissertation research on human target identification technology and proposes a human target identification algorithm based on mobile dual-mode radar platform.By using the drone-based hybrid frequency modulated continuous wave(FMCW)radar system,the algorithm can first merge different SAR images from multiple trajectories in the FMCW mode,and then identify the vital sign signal using the single-frequency CW mode.Finally,the 2-D position of human subjects can be extracted accurately.The experimental results demonstrate that compared with the traditional SAR imaging results,the proposed algorithm can not only perform SAR imaging on a wide range of scenes,but also achieve vital sign detection and single human target identification.(3)To overcome the problems of high detection complexity and multi-target recognition in the human target identification algorithm based on mobile dual-mode radar system.This dissertation proposes a human target automatic identification algorithm with a mobile single-mode radar platform.The algorithm uses a single-mode FMCW radar,which greatly simplifies the hardware and signal processing steps.It can achieve the automatic detection of human subjects by analyzing their breathing signals.The experimental results show that compared with the human target identification algorithm,the proposed method can not only identify multiple human targets in 2-D imaging and identify human targets in different complex scenes,but also extract their complete breathing waveforms,showing great potential for urban rescue applications.(4)To investigate the breathing identification of human subjects.This dissertation conducts an in-depth research on the breathing features of different human subjects and presents a human target identity recognition method based on breathing features.By fully considering the difference of breathing signals between human targets,the proposed method obtains continuous breathing signals of multi-target in the training set,and extracts 16 effective features from the radar-detected breathing waveform and establishes a training model.The experimental result shows the method can recognize specific targets only by extracting and analyzing breathing signals under supervised learning.The accuracy of 93.3% further illustrates the stability and effectiveness of the proposed method.