Parameter Estimation And Modeling Of Walking Human Based On Micro-Doppler Signature

Radar can detect human targets at a long distance.Compared with other tools,radar has many irreplaceable advantages.The human movement includes not only the uniform motion of the body,but also the non-rigid motion of the limbs and the trunk,which is very complicated.Using the Doppler Effect,the radar can accurately detect these translation and micro movements of the human body,and accurately reflect the motion information in the micro-Doppler spectrum.This thesis focuses on the human walking model.By analyzing the characteristics of micro-Doppler in the radar echo of human walking,the typical human target parameters are estimated,which can provide useful features for human body target recognition.This paper also studies the improvement method of human walking model,which has a good reference for human body modeling.The work and the main research results in this thesis include:First,the radar echo modeling method of walking human target is studied.Firstly,the Boulic global human walking model is introduced.According to its basic principle,the modeling of human walking is realized.Then,the radar echo modeling method for any relative motion of people and radar is proposed.Finally,the radar echo signal of human walking is established,and based on the linear frequency modulation signal,the basic signal processing such as pulse compression,time-frequency analysis are performed to the radar echo and the simulation results are given,which lays a theoretical foundation for the research of human model parameter estimation and modeling.Secondly,the estimation method of the global parameter of human body is studied.Based on the standardization formula of human walking time and space characteristics proposed by Inman,the influence of different height and relative velocity parameters on the micro-Doppler spectrum of human walking was analyzed and the estimation method of human global variable parameters based on the key frequency component was proposed.This method proposes the empirical energy threshold of the separation of the key Doppler frequency components.Using the key frequency component information,the problem that the human model does not consider the individual walking characteristics and does not match the actual human walking is solved,so as to achieve the accurate estimation of the human model parameters such as height and relative speed.The method is simple and the operation is small.Besides,the requirements for radar working frequency are not high.The experimental results show that the method can accurately estimate the relative speed and height of human walking at the relative velocity resolution of 0.1 m/s and height resolution of 0.1 m.Thirdly,the improved method of human body model based on global variable parameters is studied.Based on the influence of the movement of different human parts on the micro-Doppler spectrum,through comparing the micro-Doppler spectrum and the simulation data of the measured data under the same global variable parameters,the model improvement method of adjusting the independent Non-global variable is proposed to solve the problem of the poor matching of the trunk movement,and the model improvement method of adjusting the radar backscatter cross-sectional area is proposed to solve the problem of the poor match of the movement of thigh region.The experimental results show that the correlation coefficient between simulation data and measured data is obviously improved after improving the human model.Finally the improved human walking model is applied to the estimation of human height and relative velocity.The experimental results show that the proposed method still can estimate the relative velocity and height of walking human accurately under the relative speed resolution of 0.1 m/s and height resolution of 0.1 m,and it improves the estimation resolution of height.