摘要
根据合成孔径雷达图像的特点,提出一种基于SIFT特征的合成孔径雷达(SAR)景象匹配的方法。首先利用改进的特征描述符初步提取实时图与参考图的SIFT关键点;然后利用距离比和RANSAC算法去除错配,匹配出可靠的同名点对;最后计算反映实时图和参考图之间变换关系的转换参数,完成景象匹配。实验结果表明,本方法快速实用,有较强的有效性和鲁棒性。
According to the characteristic of Synthetic Aperture Radar （SAR） images, a new scene matching algorithm for SAR images based on Scale Invariant Feature Transform （SIFF） features was proposed. Firstly, SIFT keypoints of realtime images and reference images were extracted efficiently using improved SIFT feature descriptor. Then ratio of distance and Random Sample Consensus （RANSAC） were performed to guarantee the stability of matching points. Finally, transformation parameters between realtime and reference images were computed by reliable matching points. The experimental results demonstrate that our approach is fast, reliable, efficient and robust.
出处
《计算机应用》
CSCD
北大核心
2008年第9期2404-2406,共3页
journal of Computer Applications
基金
航天基金资助项目(0747-0540SITC2099-4)
关键词
SIFF特征
合成孔径雷达
景象匹配
特征描述符
Scale Invarlant Feature Transform （SIFT） feature
Synthetic Aperture Radar （SAR）
scene matching
feature descriptor
