一种低秩稀疏矩阵分解联合目标轨迹区域提取的ViSAR-GMTI方法  

A ViSAR-GMTI algorithm based on low-rank sparse decomposition and target trajectory region extraction

作　　者：尹钟政 任雨薇 郑明洁 YIN Zhongzheng;REN Yuwei;ZHENG Mingjie(Aerospace Information Research Institute,Chinese Academy of Sciences,Beijing 100190,China;School of Electronic,Electrical and Communication Engineering,University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院空天信息创新研究院,北京100190 [2]中国科学院大学电子电气与通信工程学院,北京100049

出　　处：《中国科学院大学学报（中英文）》2024年第4期533-540,共8页Journal of University of Chinese Academy of Sciences

基　　金：国家自然科学基金(61971401)资助。

摘　　要：视频合成孔径雷达(ViSAR)中,由于观测角度的不同会引起物体后向散射系数变化从而产生动态背景,不利于复杂场景下的运动目标检测,故提出一种基于低秩稀疏矩阵分解和目标运动轨迹区域提取的ViSAR运动目标检测方法。首先,考虑目标的空间连续性以及复杂场景下的诸多干扰因素,对常规RPCA模型做出改进,引入结构稀疏范数和动态背景鲁棒项以完善分解模型,提升分解效果。然后,改进局部自适应阈值设定,使用复合聚类分割方式提取运动轨迹区域,进一步消除干扰,并在分解所得前景图像的轨迹区域上进行均值背景建模完成运动目标检测。最后,基于齐鲁一号的聚束数据进行实验,结果证明所提方法的有效性,并通过对比实验验证该方法的检测性能。Different observation angles cause changes in the backscattering coefficients of objects resulting in dynamic backgrounds in video synthetic aperture radar(ViSAR),which is not conducive to the detection of moving objects in complex scenes.A ViSAR moving target detection method based on low-rank sparse decomposition and motion trajectory region extraction is proposed.First,considering the spatial continuity of the target and many interference factors in complex scenes,the conventional RPCA model is improved,and the structured sparsity-inducing norm and robust structure for dynamic background are applied in the model to obtain a better decomposition effect.Secondly,the setting of the local adaptive threshold is optimized,and the composite segmentation method is used to extract the motion trajectory area to further eliminate the interference.The mean background modeling method is used to complete the moving object detection in the trajectory area of the foreground image.Finally,the experimental results based on Qilu-1 data show the effectiveness of the proposed method,and the detection performance of the method is verified by comparative experiments.

关 键 词：视频SAR 运动目标检测 低秩稀疏分解 阈值分割 

分 类 号：TN957.52[电子电信—信号与信息处理]