摘要
本文针对较大旋转角度影像的匹配问题,提出一种基于SIFT特征匹配旋转补偿的影像匹配方法。即采用SIFT特征匹配方法估计影像间的旋转角度,对匹配窗口影像进行旋转补偿,最后采用带核线几何约束条件的相关系数法寻找同名点。在每层金字塔,采用带模型连接条件的相对定向来删除错误匹配点。通过对一组实际航摄影像进行试验,验证了所提出方法的有效性。
Aiming at the aerial digital imagery matching with large rotation, this paper proposed a matching method based on SIFT feature matching and rotation compensation. Firstly, it used the result of SIFT feature matching method to estimate the rotation angle between the images, then made rotation compensation to the matching window and adopted the geometrically constraint cross correlation algorithm to determine the conjugate points. In each pyramid level, relative orientation with model connection condition was carried out to eliminate the wrong match points. Experiment was performed on a set of practical aerial images with large rotation angle to validate the method.
出处
《测绘科学》
CSCD
北大核心
2011年第3期19-21,共3页
Science of Surveying and Mapping
关键词
影像匹配
大旋转角
SIFT特征匹配
自动化
image matching
large rotation angle
SIFT feature matching
automation