摘要
在基于点特征的遥感图像配准过程中,特征点的自动提取和准确匹配是影响配准精度的关键。提出了一种干涉合成孔径雷达(InSAR)复图像对的自动配准算法,利用特征点的高曲率结构设计了模板,完成了特征点检测;根据匹配点对之间距离相近的结论设计了匹配算法,进行了特征点对的匹配。首先通过边缘检测和模板相关提取特征点;其次根据提出的匹配算法建立点的对应关系;最后利用两步法完成复图像的亚像元级配准。实验结果表明,该算法具有较高的配准精度。
Detecting feature points automatically and accurate feature matching are the two crucial steps in InSAR image registration. In this paper,an automatic feature-based registration algorithm is presented. Two major steps are concentrated on feature detection and feature matching. In feature detection, the templates are designed to extract feature points by utilizing the high curvature property. In feature matching the process of the feature matching is completed by using the matching algorithm proposed according to the conclusion of the near distance. With edge detection and template correlation, the feature points in both images are extracted. After the corresponding relations are established by the proposed matching algorithm, a two-step method is applied to registering with sub-pixel accuracy. Experimental result shows that the proposed method can achieve high accuracy and less computational complexity.
出处
《航空学报》
EI
CAS
CSCD
北大核心
2007年第1期161-166,共6页
Acta Aeronautica et Astronautica Sinica
基金
国家自然科学基金(60372049)
关键词
遥感信息工程
图像配准
两步法
特征点匹配
干涉合成孔径雷达
亚像元级
remote sensing information engineering
image registration
two-step method
feature points matching
InSAR
sub-pixel accuracy
