基于改进的CNN和数据增强的SAR目标识别  

SAR image target recognition based on combinatorial optimization convolutional neural network

作　　者：王彩云[1] 吴钇达 王佳宁 马璐 赵焕玥 WANG Caiyun;WU Yida;WANG Jianing;MA Lu;ZHAO Huanyue(College of Astronautics,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China;Beijing Institute of Electronic System Engineering,Beijing 100854,China)

机构地区：[1]南京航空航天大学航天学院,江苏南京210016 [2]北京电子工程总体研究所,北京100854

出　　处：《系统工程与电子技术》2022年第8期2483-2487,共5页Systems Engineering and Electronics

基　　金：国家自然科学基金(61301211);国家留学基金(201906835017)资助课题。

摘　　要：针对合成孔径雷达(synthetic aperture radar, SAR)图像目标识别问题,提出了基于改进的卷积神经网络和数据增强的SAR目标识别方法。首先在训练阶段引入Dropout,随机删除部分神经元,增强网络的泛化能力。其次,在网络中引入L2正则化,简化模型的同时降低结构风险,并且能有效地抑制过拟合。然后,采用Adam优化网络,提高模型的收敛效率。最后,采用优选的数据增强方法,扩充SAR目标数据集,为网络训练提供更为充足的样本,进一步提高识别的准确率和模型的泛化性。在运动和静止目标获取与识别(moving and stationary target acquisition and recognition, MSTAR)数据集上进行了实验,结果表明设计的卷积神经网络识别准确率高,且具有更好的泛化性。For the problem of target recognition in synthetic aperture radar(SAR) image, a method of SAR target recognition based on improved convolution neural network(CNN) and data augmentation is proposed. Firstly, Dropout is brought in the training phase to randomly delete some neurons, so that the generalization ability of the network is enhanced. Secondly, L2 regularization is introduced into the network to reduce the structural risk and effectively restrain the over fitting. Then, Adam is used to optimize the network to improve the convergence efficiency of the model. Finally, the preferred rotation data augmentation method is employed for expanding the data set of SAR target. Through the improved network and increased data, the recognition accuracy and generalization of the model are enhanced. Experiments on moving and stationary target acquisition and recognition(MSTAR) data set show that the proposed method has higher recognition accuracy and better generalization.

关 键 词：雷达目标识别 合成孔径雷达 卷积神经网络 数据增强 正则化 

分 类 号：TN957.52[电子电信—信号与信息处理]