Gabor原子网络法在雷达目标高分辨距离像识别中的应用

针对飞机目标的分类问题 ,介绍了一种称为 Gabor原子网络的高分辨雷达目标距离像识别算法。 Gabor原子网络的输入层采用 Gabor原子变换作为预处理单元 ,完成对特征的提取。Gabor原子网络的隐层和输出层组成一个多层前馈网络 ,采用改进的反向传播算法对权值进行调整。文中同时给出了网络在训练过程中自动调整 Gabor原子节点的特征参数的算法。对 3种缩比模型飞机的微波暗室转台数据进行了分类 ,结果表明三维空间内的 Gabor原子网络方法比一维空间内的原始距离像或 Fourier幅度方法和二维空间内的 Gabor变换或小波变换方法更适合高分辨雷达目标距离像的识别

Gabor-atom network based target recognition using high-resolution radar range profiles

A gabor atom neural networks (GANN) approach is proposed for radar target recognition using high resolution range profiles. Gabor atom transforms are used as reprocessing units in the GANN input layer. The hidden layer and the output layer form a multilayer feedforward network and their weights are adjusted using the modified backpropagation (BP) algorithm. The feature parameters of Gabor Atom nodes are simultaneously adjusted by the network. Simulations are presented to classify microwave anechoic chamber data for three different scaled airplane models. The results show that the GANN approach which is a 3 D method is more effective for recognition of high resolution radar target range profiles than other algorithms which are 1 D or 2 D methods.