SAR图像舰船目标检测的Gamma流形方法  

Ship Detection in SAR Images Using Gamma Manifold

作　　者：王琦[1] 陈彦来[1] Wang Qi;Chen Yanlai(The research Institute No.723 of CSIC,Yangzhou,Jiangsu 225101)

机构地区：[1]中国船舶重工集团公司第七二三研究所,江苏扬州225101

出　　处：《火控雷达技术》2018年第2期14-17,共4页Fire Control Radar Technology

摘　　要：目前基于统计特征的合成孔径雷达(Synthetic Aperture Radar,SAR)图像目标检测方法,易受成像条件稳定性的影响而误差较大。针对这一问题,本文提出了一种基于Gamma流形的SAR图像舰船目标检测算法。通过Gamma分布族构建SAR图像二维统计流形,构造黎曼度量,借助切向量长度实现对SAR图像的显著性表示。最后利用分类算法,实现SAR图像舰船目标的检测过程。实验结果表明,该方法不仅能有效地检测出SAR图像感兴趣目标区域,而且对海杂波具有较好的鲁棒性。Statistical features based target detection through synthetic aperture radar(SAR)imaging is usually limited by imaging stability,and which will lead to bigger error.A new method for ship detection in SAR images by using Gamma manifold is proposed based on target inherent geometrical structures.To build SAR image 2D statistic manifold through Gamma distribution family;to fabricate Riemannian measurement and to implement significant representation of SAR image with the aid of tangent vector length;and finally to achieve ship detection through SAR image by using classification algorithm.Experimental results demonstrate that the proposed method can effectively detect interested zone in SAR image,and has good robustness to sea clutters.

关 键 词：目标检测 SAR图像 信息几何 流形 

分 类 号：TN957.52[电子电信—信号与信息处理]