Research On Human Breathing Recognition Technology Based On Vital Signs SAR Imaging

Imaging and breathing recognition of human targets have extensive requirements in counter-terrorism rescue,user authentication,and health monitoring.At present,the imaging of human targets mainly includes optical imaging,infrared imaging and radar imaging.Among them,optical imaging cannot be all-weather,all-day imaging,and it will cause privacy issues.Infrared imaging is susceptible to interference from other heat sources and affects the imaging effect.The use of radar imaging in counter-terrorism rescue often makes it difficult to carry out the rescue of hostage targets due to the failure to identify the identity of human targets.Therefore,it is of great practical significance to design a system that can locate human targets and identify human target breathing.This paper presents a human respiratory recognition system based on vital sign SAR imaging.The system first establishes a breathing recognition model,secondly uses a vital sign SAR imaging algorithm to locate a human target and obtain its breathing waveform,and finally inputs the breathing waveform of the human target to be recognized into the breathing recognition model to accurately identify the human target in the scene.Its main work is as follows:1.The frequency-modulated continuous wave radar system is introduced and the detailed theoretical analysis of radar echo signals is introduced.Two classic SAR imaging algorithms based on FMCW system are introduced,and the advantages and disadvantages of the two algorithms are compared through the simulation.2.A human target positioning algorithm based on vital sign SAR imaging is proposed,which can distinguish between human targets and non-living targets.Through distance correction and phase signal processing,micro-Doppler of human targets can be extracted from a single range door.Information to determine the exact position of the human target in the SAR imaging map and verify the effectiveness of the algorithm through simulation.3.The human breathing recognition algorithm based on machine learning is introduced.Feature extraction and feature selection are performed on breathing signals.Three classic machine learning algorithms: K-nearest neighbor algorithm,support vector machine algorithm and decision tree algorithm are studied.4.Set up an FMCW radar system experimental platform.First,perform human target positioning experiments based on vital sign SAR imaging.The experimental results show that the minimum error of human target positioning is 0.05 m.Then,a human respiratory recognition experiment based on machine learning is performed.The experimental results show that the Liner SVM classification model has the highest accuracy rate for respiratory recognition,and the classification accuracy rate for 20 people is 91.1%.Finally,a human breathing recognition experiment based on vital sign SAR imaging was performed.The experimental results show that the system can accurately locate three human targets in the scene with a maximum error of0.05 m,and the accuracy of breathing recognition is 84%,92%,82% The experimental results verify the effectiveness of the system in this paper.