基于OMP-CADMM的非合作运动人体ViSAR成像方法

A ViSAR Imaging Method for Non-Cooperative Moving Human Based on OMP-CADMM

合成孔径雷达(Synthetic Aperture Radar,SAR)人体成像技术凭借其高分辨、强穿透、易部署和高安全性,已成为SAR拓展至日常民用领域的关键技术之一.然而现有SAR人体成像方法大多针对固定场景下的静态人体目标进行合作式成像,难以满足开放场景非合作成像应用的需求.为此,本文基于人体运动的块稀疏性与一致性约束,提出了一种面向非合作运动人体的视频SAR(Video SAR,ViSAR)成像方法 .该方法首先结合ViSAR子孔径快速成像技术,简化人体非刚性运动模型,建立非刚性运动相位误差参量模型;随后结合人体运动的块稀疏性与时间连续性,基于正交匹配追踪(Orthogonal Matching Pursuit,OMP)稀疏恢复算法估计人体运动相位误差;最后基于一致性交替方向乘子法(Consensus Alternating Direction Multiplier Method,CADMM)优化框架,保留人体目标局部结果的一致性共有特征,从多个局部补偿结果集中优化出人体的全局聚焦成像结果,实现人体的非刚性运动补偿与全局精细优化.实验结果表明该文所提方法可实现非合作走动人体的视频化高分辨精细成像.

Synthetic aperture radar(SAR)human imaging technology has become the key technology for SAR to ex⁃pand to the daily civil field because of its high resolution,strong penetration,easy deployment,and high security.However,most of the existing SAR human imaging methods focus on the cooperative imaging of static human targets in fixed scenes,which cannot meet the needs of non-cooperative imaging applications in open scenes.Therefore,this paper proposes a vid⁃eo SAR(ViSAR)imaging method for the non-cooperative moving human body based on the block sparsity and consistency constraints of human motion.Firstly,combined with ViSAR sub-aperture fast imaging technology,the non-rigid motion model of the human body is simplified and the phase error parameter model of non-rigid motion is established.Then,com⁃bined with the block sparsity and time continuity of human motion,the phase error of human motion is estimated based on the orthogonal matching pursuit(OMP)sparse recovery algorithm.Finally,combined with the optimization framework of the consensus alternating direction multiplier method(CADMM),the non-rigid motion compensation and global fine opti⁃mization of human body are realized.Experimental results show that the proposed method can realize video high-resolution fine imaging of the non-cooperative walking human body.