Research On Physical Parameters Estimation Of Rotor And Classification Of Ground Targets Based On Low-Resolution Radar Signatures

The micro-Doppler effect reflects the moving characteristics and geometrical structures of the targets and can be regarded as a unique signature of the targets. In recent years, the micro-Doppler effect is widely introduced into the field of radar and it is a new approach for the radar automatic target recognition(ART). Since helicopter, propeller and jet aircraft have different rotors, we can categorize aircraft targets into three kinds by the utilizing micro-Doppler effects. Besides, a moving vehicle and walking person have different micro-motion structures, so we can resolve the classification problem between a moving vehicle and walking person by the utilizing micro-Doppler effects. The main work of the paper can be summarized as the following two parts:1.We analyze the disadvantages of the aircraft targets classification method based on the micro-Doppler features extraction. Meanwhile, we propose a method of estimating physical parameters of the rotor, which can be used to classify three types of aircraft targets. The method utilizes the coherent single range Doppler interferometry(CSRDI) algorithm and the Hough transform algorithm for the narrowband imaging of the rotor. Then we use the principal component analysis(PCA) to extract physical parameters of the rotor from the narrowband imaging result of the rotor. Based on simulation data and measured data, we verify the effectiveness of the proposed method.2.Based on the differences of time-frequency spectrograms between a moving vehicle and walking people, a 4-dimensional time-frequency feature vector describing the characteristics of time-frequency spectrograms is extracted from micro-Doppler signatures. Meanwhile, we theoretically demonstrate the robustness of a 3-dimensional time-frequency feature vector including in the 4-dimensional time-frequency feature vector. The experimental results based on the measured data show that the 4-dimensional time-frequency feature vector can achieve a good discriminative ability to categorize grounding moving targets into the three kinds, i.e, single person, many person and a vehicle, compared with some existing features. Then the experimental results based on the measured data show that the 3-dimensional time-frequency feature vector can achieve a satisfactory classification performance to discriminate the moving vehicle and the walking person, compared with some existing features, when the test SNR is relatively low. Besides,the target classification process based on the 3-dimensional time-frequency feature vector has small computation time, compared with that based on some existing features with a noise reduction method.