Research On Extraction Method Of Micro-Doppler Curve For Radar Target

In the detection of radar targets,in addition to the translation of the target centroid,there are often micro-motions of the target itself or its partial structure,including small movements such as vibration,rotation,tumbling or precession.This phenomenon is called the microDoppler effect of the radar target.The micro-Doppler feature of the radar target is a unique feature of the target motion,which provides a fine description of the motion of the target.Therefore,since the micro-Doppler effect of the radar target was discovered,the micromotion characteristics of the effect target is widely used in the fields of identification,detection and classification of the radar target.Nowadays,the detection,separation,parameter estimation and feature extraction of radar target micro-motion signal have become a research hotspot in radar academic circles.For the extraction of micro-Doppler curves,there have been many research results in recent years,but the existing extraction method of Micro-Doppler curve for the radar target,including Hough transform and inverse Radon transform,have the problem of a large amount of computation.Besides,under the condition of low SNR,how to extract the micro-motion feature of the weak radar target has been a difficult problem in research for a long time.The traditional extraction method of MicroDoppler curve and parameter estimation algorithms based on time-frequency analysis usually can not get good results in the case of low SNR.In this paper,we have some research on the micro-motion characteristics and the extraction method of Micro-Doppler curve of the radar target.The content mainly includes the following three parts:Firstly,the domestic and international research status of the micro-Doppler and predetection tracking technology,the research background and main work of this paper are introduced.And then,we expound the concepts of Doppler effect and micro-Doppler effect.Next,we introduce the theory of the EM scattering of moving objects,and establish the mathematical model of micro-Doppler effect of the basic forms of micro-motion such as vibration and rotation.Secondly,we do some research on the extraction method of micro-Doppler features,including two basic time-frequency analysis methods——short-time Fourier transform and Wigner-Ville distribution.Then the micro-Doppler parameter estimation method based on inverse Radon transform and its principle and algorithm process are introduced.In the radar signal processing,the fast reflection,vibration or oscillation of the partially reflected signal causes the micro-Doppler effect in the form of a sinusoidal FM signal,so in this paper,the micro-Doppler curve is simplified to the form of sinusoid for simulation analysis.The inverse Radon transform method is proved to be a method for estimating parameter and decomposing the sinusoidal FM signal accurately and effectively,and can be used for the extraction of the micro-Doppler curve for multiple components.Finally,for the low SNR condition,some extraction methods of micro-Doppler curves for radar target are less effective,and the existing multi-component micro-Doppler feature extraction algorithms have large computational problems,such as inverse Radon transform.In order to solve these problems,this paper proposes a new extraction method of MicroDoppler curve using a multi-objective tracking algorithm with a low SNR condition in the pre-detection tracking technique——histogram probabilistic multi-hypothesis tracking algorithm.In the research process,the algorithm is firstly used to track and simulate the target of uniform linear motion,and then expands the linear target to the nonlinear target.The nonlinear observation model is established by the time-frequency diagram of the microDoppler curve,interpret the time-frequency diagram as the observation histogram of the stationary stochastic process,and use the discrete polynomial distribution to model the observation points in the entire time-frequency diagram,and probabilistically assign the total number of observation points in all resolution units to the target and the noise models,and then use the EM algorithm to achieve the maximum likelihood estimation of multi-objective model parameters,and finally achieve the separation of multiple micro-Doppler curves.To facilitate the study of this algorithm,we also simplified the model of the micro-Doppler curve into the form of intersecting sinusoids.This algorithm is proved to be a method that can extract multiple curves at the same time and effectively reduce the computational complexity.