基于分数布朗特征的低分辨雷达目标分类

Target Classification of Low-resolution Radar Based on Fractional Brown Feature

低分辨雷达目标的JEM效应不仅体现在对目标回波的频率调制上,还体现在幅度调制上,以上2种调试特征是互补的关系,综合运用2种调制特征将会改善低分辨雷达对飞机目标的分类与识别能力。分数布朗运动模型是一种可以用来描述自然界中随机分形的统计模型,文中提出将分数布朗运动模型用于反映回波幅度的不规则程度或起伏程度,以及回波幅度起伏的快慢,分别提取回波时域和频域的分数布朗分形维,再采用支持向量机分类器进行分类,很好地将直升机与民航机2类目标区分开来。实测数据表明,综合分类正确率大于85!。

The JEM （Jet Engine Modulation）effects of target in Low-resolution Radar are existed not .only in the frequency modulation, but also in amplitude modulation of the target echo. These two modulation characteristics are independently, using them compositively will improve the ability of airplane target classification and recognition of Low-resolution. Fractional Brown motion module is a statistical module using to describe the random fractal. This paper purpose to use fractional Brown motion module to express the degree of fluctuation or abnormity of target amplitude, and the speed of the fluctuation of the target amplitude, extract the fractional Brown motion fractal dimension of the echo in time and frequency domain separately, by using SVM classificator, the helicopter and civil airplane can be distinguished primly. The real recorded echo data are processed and the results show that the correct classification rate is greater than 85 %.