Microdoppler Feature Extraction In Time-frequency Radar Imaging

The micro-doppler effect can reflect a series of information such as the detailed characteristics and motion details of the target.However,how to separate the target signal from the micro-doppler information to extract the micro-doppler features and how to make the radar imaging clearer is also a hot issue in the field of micro-doppler feature extraction.Because the time-frequency representation of each distance gate has obvious multi-component time-frequency characteristics,the instantaneous frequency estimation is an ideal tool for micro-doppler feature extraction.Instantaneous frequency is an important feature of the non-stationary signal,which is instantaneously effective.It can reflect the frequency change trend of the signal at a certain time and represent the local characteristics of the time-frequency.However,the instantaneous frequency estimation of multi-component signals has always been a difficult problem in the field of non-stationary signal processing,especially the instantaneous frequency estimation of overlapping multi-component signals.In this thesis,the estimation problem of multi-component instantaneous frequency is solved.On this basis,the problem of micro-doppler feature extraction in radar imaging has been solved.Based on the short-time Fourier transform in time-frequency analysis and RANSAC algorithm,an algorithm for instantaneous frequency estimation and separation of multi-component overlapped signals is proposed.When S transform is applied to radar imaging,combining with the algorithm previously proposed,the target and micro-doppler information are separated to extract micro-doppler features.The main work of this thesis includes:1.An adaptive RANSAC algorithm is proposed to estimate the instantaneous frequency of multi-component overlapping signals.This algorithm combines the parametric method and the non-parametric method,which does not need to know the prior knowledge such as the number of signal components.In addition,a de-noising method is added to make the algorithm work effectively at low SNR.Experimental results show that multi-component overlapping signals can be separated accurately under low SNR.2.To solve the problem of fuzzy target imaging based on Fourier transform,S transform and generalized S transform are applied to radar imaging.In the time-frequency analysis plane,the target signal and the micro-doppler information part are the single frequency signal of the low frequency part and the sine frequency modulation signal of the high frequency part respectively.Due to the characteristic of S transformation to high frequency diffusion,matching radar imaging can reduce the interference of micro-doppler component in the target signal part of low frequency part,which is conducive to the separation of target signal and micro-doppler component.3.The Singular-Spectrum Analysis(SSA)is used to separate the target signal from the micro-doppler component.The method of singular value decomposition is used to decompose the non-rigid target echo signal,and the echo of the rigid body part of the target is separated from the micro-doppler information introduced by the moving parts.Simulation experiments and actual data show that the algorithm can effectively separate the micro-doppler features.4.The RANSAC algorithm is used to extract the micro-doppler features,and the experimental results showed that the algorithm could effectively separate the rigid body part and the micro-doppler information part in the radar echo signal,and the experimental results of the two algorithms were compared and analyzed.