雷达目标双谱特征分析与分类方法研究

Study of bispectrum signatures of radar targets andclassification techniques

利用复信号双谱估计技术分析了常规相参雷达实测飞机目标回波信号，研究指出目标回波是典型的非高斯信号，且其双谱分布区域信息较好地反映了目标的部分散射调制特性，可作为目标分类特征．在此基础上，提出一种目标双谱特征抽取和模板构造方法．利用双谱经阈值化处理后的二值特征图，分别定义了“核”和“域”来表征目标双谱分布区域的共性和范围，并用其构造目标模板．实测数据实验结果表明了该方法在分类效果和特征降维方面的有效性．

A bispectrum analysis of real target echoes based on conventional coherent radar is made using the complex bispecturm estimation. The useful facts are revealed that the target echoes are typical non Gaussian signals and that the bispectrum distribution domain information effectively reflects the partial scattering modulation signature of targets, which can be used as the signature for target classification. On the basis of this, a method is proposed for bispectrum signature extraction and template construction. By making use of the binary signature image created by thresholding the bispectrum, a “Core” and a “Domain” are defined to represent the generality of bispectrum and its distribution domain, with a template being constructed by them. Finally, the effectiveness of this method in both classification and signature dimensionality reduction is illustrated with some experiments on real target echoes.