Stable and effective target feature extraction is very significant for target classification of low-res-olution radar.An approach to classify the radar target feature from the radar cross section (RCS)and spectral entropy probability distribution curves is proposed.The target spectrum calculation by the fast Fourier trans-form (FFT)is first used to obtain the target RCS and spectrum entropy.Then the probability distribution curves of the part basic features are calculated,and the stable target feature is extracted from the curves.The measured data results based on the support vector machine show the proposed feature can not only be robust to the target and achieve good classification performance,but also implement simply.%提取稳定有效的目标特征对于低分辨雷达的目标识别分类有着重要意义。在提取目标基本特征雷达散射截面积(radar cross section,RCS)与频谱熵值的基础上,提出了一种基于特征概率分布曲线的目标分类方法。该方法首先应用快速傅里叶变换计算目标回波频谱,提取目标 RCS 与频谱熵值,然后滑窗分段计算基本特征的概率分布曲线,从而利用概率分布曲线提取出稳定的目标特征,最后利用支持向量机对目标实现分类。基于实测数据的分类结果表明,该特征具有较好的稳健性和分类性能,同时算法便于工程实现。
展开▼