超宽带探地雷达自动目标识别研究

Automatic target recognition for ultra-wideband ground-penetrating radar

超宽带探地雷达由于回波信号信息丰富的特点,特征向量的选取成为了自动目标识别的关键。首先利用宽相关处理进行滤波和典型数据提取。提取纵向和横向典型数据用于目标形状识别;提取典型回波道数据进行功率谱分析用于目标材质识别,并利用RBF网络从而实现目标的自动识别。实测数据处理表明:该方法可有效的完成管状物和球状物、铝和铁的识别。

Because of abundant information in the signal echo of ultra-wideband ground-penetrating radar, Signature extraction becomes the key of automatic target recognition. In this paper, the wideband correlation processing is empolyed to filter and extract the typical data. The shape of target is recognized by the longitudinal and transverse typical data, and the composition of target is classificated by the power spectrum of the typical echo, and the automatic target recognition is implemented by using the RBF. The processing results of measured data indicate that the present method can effectively distinguish tube and ball, iron and aluminium.