Modeling And Feature Extraction Of Moving Human Target In Millimeter Wave Radar

The classification of human motion state has always been one of the problems in the radar field due to the complex motion state of the human body,the variable scattering characteristics,and the strong clutter interference in the detection environment.This paper focuses on the key techniques of detection mechanism of moving human targets,pedestrian echo model,motion feature extraction,motion state classification,and measured data processing.Firstly,this paper establishes a simplified human motion model based on the Boulic model.We convert the walking process of the human body into a combination of rigid body translation and rotational motion,and convert the body parts into elliptical rigid bodies.We have established a human motion echo model of the LFMCW radar.We verified the correctness of the model by comparing the time-frequency plots of the model data with the measured data.Secondly,Traditional time-frequency analysis method has poor time-frequency focusing.This paper betterly describes the human micro-Doppler by combining multisynchrosqueezing transform and time-frequency analysis.We implemented frequency compression in the time-frequency spectrum.In this paper,Rényi entropy is used as the criterion for the time-frequency spectrum energy focus,and the criterion for multisynchrosqueezing transform threshold determination.Furthermore,for the problem of the diffusion of the LFMCW radar range image,this paper designs a diffusion image detection algorithm based on time-frequency spectrum in order to make full use of micro-Doppler information of human targets.This paper extracts human motion features through image edge detection.We classify the motion state by SVM classifier.Finally,we verify the feasibility of the algorithm by the measured data obtained of millimeter wave radar.