基于卷积神经网络的雷达目标HRRP分类识别方法

卷积神经网络通过卷积和池化操作提取图像在各个层次上的特征进而对目标进行有效识别,是深度学习网络中应用最广泛的一种。文中围绕一维距离像雷达导引头自动目标识别,开展基于卷积神经网络的目标高分辨距离像分类识别方法研究。首先,基于空中目标一维距离像姿态敏感性仿真生成近似平行交会条件下不同类型目标的高分辨距离像数据集;其次,构建一种一维卷积神经网络结构对目标高分辨距离像进行分类识别;作为比较,针对同类高分辨距离像数据集,分析了主成分分析-支持向量机方法的目标分类识别效果。结果表明:基于卷积神经网络的目标分类识别算法有更好的识别能力,对高分辨距离像的姿态敏感性具有较强的适应性。

HRRP Classification and Recognition Method of Radar Target Based on Convolutional Neural Network

Convolutional neural network(CNN) is a deep learning network which has been widely used. CNN has a good performance on target classification and recognition. Features of images can be extracted by CNN through convolution and pooling operations. A radar automatic target recognition algorithm has been discussed in this paper. It is based on high range resoluton profile (HRRP) and CNN. Firstly, HRRPs of several air targets under approximately parallel intersection conditions have been generated to construct the dataset. Then, a CNN has been used to classify and recognize the HRRPs. Finally, the result with another algorithm which is based on principal component analysis and support vector machine are compured, that CNN has a better performance and strong adaptability to the attitude sensitivity of HRRPs.