机载雷达具有速度快、视野广的特点,适用于战场运动目标的跟踪识别。而目前国内外针对机载低分辨雷达下微动地面目标识别的研究尚属空白。文中建立了机载平台下战场目标的雷达回波散射点模型,分析了单兵、轮式车、履带车等典型目标的微动特性,根据行进单兵微多普勒的周期性特征实现了单兵与机动车的识别;进而通过提取回波信号特征谱和支持向量机的方法实现轮式车与履带车的分类,在20 dB信噪比下识别率在87.5%以上。分类方法具有稳健性和有效性。

Airborne radar is suitable for battlefield targets tracking and recognition with its high speed and wide view. While little work has been done on ground targets classification with micro-motion in low-resolution airborne radar. In this paper, point-scattering echo models are established for moving targets, micro-Doppler signature of typical targets is analyzed, and recognition of singlesoldier and vehicle is achieved according to the unique periodicity of single-soldier's micro-Doppler signature. Eigenvalue spectrum is selected as a feature for classification and support vector machine is used for classification of wheeled and tracked vehicles. A recognition rate of over 87.5% is achieved with SNR of 20 dB. Experimental results validate that the selected feature is robust and the classification method is effective.

TN959.73

国家863基金资助项目