一种基于双向匹配融合的合成孔径雷达景象匹配算法  

A SAR Scene Matching Algorithm Based on Bi-directional Matching Fusion

作　　者：伍薇 苏娟 WU Wei;SU Juan(Rocket Force University of Engineering, Xi an 710025, Shaanxi, China)

机构地区：[1]火箭军工程大学,陕西西安710025

出　　处：《兵工学报》2019年第1期124-133,共10页Acta Armamentarii

基　　金：国家自然科学基金项目(61302195)

摘　　要：针对合成孔径雷达(SAR)图像进行景象匹配时因图像差异和斑点噪声造成误匹配的问题,提出了一种基于双向匹配融合的景象匹配算法。采用变形差异相似性测度进行从实时图到参考图的正向匹配;通过选取参考图中易匹配子区,进行基于相位一致性特征强度互相关的从参考图到实时图的反向匹配;对双向匹配的结果进行融合处理,确定最终的匹配位置。实验结果表明,该算法能克服异源SAR图像间的图像差异对匹配的影响,具有较高的匹配精度和较好的鲁棒性。A SAR scene matching algorithm based on bi-directional matching fusion is proposed.The deformable diversity similarity is used to realize the forward matching from the real-time image to the reference image.The subregions which are easy to be matched are chosen in the reference image,and the normalized cross correlation of phase congruency feature is adopted to realize the reverse matching from the reference image to the real-time image.The final matching location is determined by fusing the bi-directional matching results.Experimental results show that the proposed algorithm has high accuracy and strong robustness,which can overcome the influence of image difference on matching.

关 键 词：合成孔径雷达图像 双向匹配 变形差异相似性 相位一致性特征 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]