基于微多普勒特征的直升机目标识别

Helicopter Target Recognition Based on Micro-Doppler Feature

微多普勒特征可为旋翼直升机的识别提供一种新的依据。现有针对旋翼直升机回波模型的研究仅限于矩形旋翼叶片尖端的微多普勒效应。首先,通过建立矩形尖端旋翼、锥形尖端旋翼以及抛物线扫掠尖端旋翼的全散射点回波模型,引入特征提取算子,设计“时频分析—归一化时频处理—频率‘投影压缩’—截断取模傅里叶变换”的方法,实现具有时移幅频不变特性的微动特征的提取。然后,基于交替方向乘子法,提出在目标分类中的压缩感知问题求解算法,并与传统的支持向量机方法进行对比分析,借助多个仿真实验验证所提出方法的有效性。

Micro-Doppler features provide an inventive basis for identifying helicopters'rotors.The existing research on the echo model of helicopters'rotors is limited to the micro-Doppler effect caused by the tips of rectangular rotor blades.This study has initially established the full-scatterer echo models of rectangular tip rotor,tapered tip rotor,and parabolic swept tip rotor.Then,utilizing the feature extraction operator,a method of"time-frequency analysis(TFA)-normalized time-frequency processing-frequency'projection and compression'-'truncation and amplitude calculation'Fourier transform"was adapted to extract micro-Doppler features.It could effectively translate in the time domain while constantly maintaining the amplitude-frequency characteristics.Finally,an algorithm solving compressed sensing problems in target classification was proposed using the alternating direction method of multipliers(ADMM)and was compared with traditional support vector machines.The effectiveness of the proposed approach was acquired from simulation results.