摘要
在雷达探测领域,数据样本无论在完备性还是多样性上,均不能满足深度学习模型有效训练的要求,模型极易出现过拟合现象,从而限制了相关技术在雷达探测领域的广泛应用。面向雷达探测领域的智能化应用需求,针对雷达数据样本不足问题,提出基于生成对抗神经网络的微波成像体制雷达数据增广方法。针对雷达数据样本特征不显著问题,结合标签平滑正则化方法,实现增广数据样本的自动标注,通过构建增广样本与真实样本协同的深度学习模型训练框架,实现模型在小规模雷达数据样本集上的鲁棒训练。基于公开雷达探测数据集,验证了该方法的有效性。
In the research field of radar remote sensing,both the completeness and diversity of radar data samples cannot meet the requirement of effective training of deep learning models,and the models are prone to over-fitting,which significantly limits the wide application of deep learning techniques in this field.Targeting on the needs of intelligent application in radar remote sensing,a microwave imaging radar suited data augmentation method is proposed to solve the issue of insufficient radar data samples by leveraging the general framework of generative adversarial network.Aiming at the features of radar samples being not obvious,the label smoothing regularization technique is utilized to automatically classify the augmentated radar samples.The augmentated samples together with the real samples are collaboratively used to implement the robust training of deep learning models.The proposed method is verified by the experiments based on the extensive open-sourse radar remote sensing data.
作者
康旭
张晓峰
Kang Xu;Zhang Xiaofeng(Beijing Institute of Remote Sensing Equipment,Beijing 100854,China)
出处
《系统仿真学报》
CAS
CSCD
北大核心
2022年第4期920-927,共8页
Journal of System Simulation
关键词
深度学习
雷达遥感探测
生成对抗神经网络
数据增广
deep learning
radar remote sensing
generative adversarial network
data augmentation