基于局部双谱的高分辨距离像雷达目标识别

针对飞机目标的分类问题 ,介绍了一种基于局部双谱的高分辨雷达目标距离像识别方法 ,利用 Fisher类别可分性鉴别测度在距离像的双谱平面选择用于分类的局部双谱。将局部双谱方法与现有的几种基于双谱的方法进行比较 ,分析出局部双谱既利用了双谱的平移不变性 ,实现了一维匹配 ,又避免了在识别过程中计算整个双谱 ,减少了计算量 ,而且不需考虑极坐标双谱的插值与积分步长问题。利用三种缩比模型飞机的微波暗室转台数据对局部双谱的分类性能进行了测试 ,结果表明局部双谱是一种很有价值的平移不变特征 ,对于高分辨雷达目标距离像具有较高的识别率

Local bispectra-based high-resolution radar target recognition with range profiles

Shift invariant features, the local bispectrum of range profiles, are proposed for application of high resolution radar target recognition. Local bispectrum is selected through the Fisher class separability discriminant measure on the bifrequency plane. The local bispectrum based classification method conserves the shift invariant characteristic of the bispectrum. The method does not calculate and use the whole bispectra for 2 D matching but achieves 1 D matching, which is more computationally economic. This method avoids the interpolation and the step problem for the integration process on the polar bispectra plane faced by other integrated bispectrum based methods. The selected features are evaluated using range profiles from microwave anechoic chamber data for three different scaled airplane models. The results are highly discriminative and provide proper recognition of high resolution radar target range profiles.