摘要
合成孔径雷达(Synthetic Aperture Radar,SAR)具有全天时、全天候、高分辨率对地观测的优势,但在成像过程中容易受到电磁干扰,进而严重影响SAR图像的判读与解译.针对复杂对抗环境下的强有源压制干扰问题,本文提出一种基于自监督复数域深度学习的SAR有源压制干扰抑制方法,以及一种新型复数域干扰抑制网络,对权值、激活函数及卷积运算等进行了复数域处理设计,挖掘SAR复数域图像中目标和干扰在幅度和相位两方面的不同信息表征,实现对干扰的抑制.同时本文提出一种自监督训练策略,解决传统网络训练过程严重依赖人工标注样本的问题,其适用于复杂干扰下样本难以标注的应用场景.开展仿真分析与实测数据验证,实验结果表明所提方法可有效地抑制复杂背景下的有源压制干扰,具有自监督智能干扰抑制能力.
Synthetic aperture radar(SAR)has the advantages of all-day,all-weather,and high-resolution in the earth observation,but it is susceptible to electromagnetic interference during the imaging process,which seriously affects the sub-sequent interpretation of SAR images.To this end,this paper proposes a suppression method for SAR blanketing jamming based on self-supervised complex-valued deep learning,and proposes a novel complex-valued interference suppression net-work,which can make full use of the amplitude and phase information of SAR complex images.The weights,activation functions and convolution operations of the network are designed for complex domain processing,and the different informa-tion representations of target and clutter in amplitude and phase in SAR images are mined to achieve interference suppres-sion.Meanwhile,a self-supervised training strategy is proposed to solve the problem of relying heavily on manually labeled samples in the traditional network training process,and is suitable for application scenarios where samples are difficult to be labeled under complex interference.The simulation analysis and experimental verification are carried out.The experimental results show that the proposed method can effectively suppress the active jamming of complex backgrounds,and has the ability of self-supervised intelligent interference suppression.
作者
化青龙
魏晨曦
张云
张倩
冀振元
姜义成
HUA Qing-ong;WEI Chen-xi;ZHANG Yun;ZHANG Qian;JI Zhen-yuan;JIANG Yi-cheng(School of Electronics and Information Engineering,Harbin Institute of Technology,Harbin,Heilongjiang 150001,China)
出处
《电子学报》
EI
CAS
CSCD
北大核心
2023年第4期965-974,共10页
Acta Electronica Sinica
基金
国家自然科学基金(No.61201304,No.61201308)。
关键词
合成孔径雷达
复数域深度学习
自监督
干扰抑制
压制性干扰
synthetic aperture radar(SAR)
complex-valued deep learning
self-supervision
interference suppres-sion
blanketing jamming
