高分雷达与光学影像融合的滨海湿地变化检测  

Coastal wetland change detection using fusion of high resolution radar and optical images

作　　者：吴瑞娟 何秀凤[2] WU Ruijuan;HE Xiufeng(School of Geography and Resource Science,Neijiang Normal University,Neijiang,Sichuan 641100,China;School of Earth Sciences and Engineering,Hohai University,Nanjing 211100,China)

机构地区：[1]内江师范学院,地理与资源科学学院,四川内江641100 [2]河海大学,地球科学与工程学院,南京211100

出　　处：《测绘科学》2020年第11期93-100,共8页Science of Surveying and Mapping

基　　金：国家自然科学基金项目(41830110,41871203);内江师范学院科研资助项目(2019YZ02)。

摘　　要：为提高滨海湿地变化检测的精度,该文以合成孔径雷达(SAR)与光学影像各自优势和互补性为切入点,研究SAR与光学影像数据融合的变化检测新方法。提出了高分SAR与光学影像特征自适应融合方法,选用江苏盐城滨海湿地TerraSAR-X、高分三号SAR影像和资源三号光学影像进行实验。研究结果表明,自适应加权的多特征融合方法优于传统多特征融合方法,虚检率明显降低。相比于传统像元级、对象级、SG-PCAK和SG-RCVA-RF变化检测方法,显著图引导的结合像元级与对象级变化检测方法提高了变化检测精度。In order to improve the accuracy of monitoring changes in coastal wetlands,taking the advantages and complementarities for describing the change of coastal wetlands of SAR(Synthetic Aperture Radar)and optical images,new change detection methods by fusing SAR and optical images were proposed.An adaptive fusion of high-resolution SAR and optical image features was proposed,in order to resolve the problem that the single-source remote sensing data could not fully express the complex change information of coastal wetlands.In this paper,Ziyuan-3 images,Gaofen-3 image and TerraSAR-X image in Yancheng coastal wetland areas were used,the experimental results demonstrated that the adaptive image features fusion approach was superior to traditional multi-feature fusion methods,and the false detection rate was significantly decreased,the proposed change detection method combining pixel-based and object-based methods improved the accuracy of change detection,compared with traditional pixel-based,object-based,SG-PCAK and SG-RCVA-RF methods.

关 键 词：遥感变化检测 特征融合 显著性检测 不确定性指数 随机森林 滨海湿地 

分 类 号：P237[天文地球—摄影测量与遥感]