摘要
提出面向合成孔径雷达(Synthetic Aperture Radar,SAR)回波数据的复杂结构特征增强算法(Complex Structure Feature Enhancement Algorithm,CEA),面向SAR成像目标的复杂结构特征,算法利用高阶方向全变分(High-order Total Direction Variation,HOTDV)正则算子表示,面向SAR成像目标的稀疏特征,算法用ℓ_(1)正则算子表示。算法利用交替方向多乘子法(Alternating Direction Method of Multipliers,ADMM)建立多正则约束优化框架,设计复杂结构分裂变量和稀疏分裂变量,并求出分裂变量解析更新解以实现SAR成像目标的复杂结构特征与稀疏特征的增强。多正则约束优化框架中的对偶分解保证多特征多任务处理能力,增广拉格朗日项的使用则保证了算法的收敛性和稳健性。最后,设计了仿真和实测SAR数据特征增强实验以验证算法的有效性,对比多种传统结构特征增强算法以验证所提复杂结构特征增强算法的优越性。
A complex structure feature enhancement(CEA)algorithm was proposed,and in which the complex struc‐ture is represented by high-order total direction variation(HOTDV)regular operator,and the sparse feature of SAR im‐aging targets is represented by ℓ_(1) regular operator.The algorithm uses Alternating Direction Method of Multipliers(AD‐MM)to establish a multi-regular constraint optimization framework,and designs complex structure splitting variables and sparse splitting variables.The analytic solutions of split variables are obtained to enhance the complex structure fea‐tures and sparse features of SAR imaging targets.The thinking of“dual-decomposition”in the multi-regular constraint optimization framework guarantees the multi-feature processing capability,while the use of Augmented Lagrange term guarantees the convergence and robustness of the algorithm.Finally,simulation and measured SAR data complex struc‐ture feature enhancement experiments are designed to verify the effectiveness of the proposed algorithm,and several tra‐ditional structure feature enhancement algorithms are compared to verify the superiority of the proposed complex struc‐tural feature enhancement algorithm.
作者
黄博
周劼
江舸
张海
HUANG Bo;ZHOU Jie;JIANG Ge;ZHANG Hai(Institute of Electronic Engineering,Chinese Academy of Engineering Physics,Mianyang 621999,China)
出处
《红外与毫米波学报》
SCIE
EI
CAS
CSCD
北大核心
2022年第4期762-769,共8页
Journal of Infrared and Millimeter Waves
基金
国家自然科学基金委员会-中国工程物理研究院NSAF联合基金(U2130202)。
关键词
合成孔径雷达
复杂结构特征增强
高阶方向全变分
交替方向多乘子法
近端算子
相位误差补偿
synthetic aperture radar(SAR)
complex structure features enhancement(CEA)
high-order total direction variation(HOTDV)
alternating direction method of multipliers(ADMM)
proximal mapping
phase error compensation
