摘要
针对非量测相机像对的定向参数精度差、不稳定等问题,本文以恢复影像的定向参数作为出发点,提出了多重约束条件下SIFT匹配的影像定向方法,首先在尺度不变SIFT算法中提取影像间同名点,为进一步剔除误匹配点,提高同名点的精度,在此基础上引入左右对称一致性、随机抽样算法、仿射变换进行逐级约束匹配,为相对定向提供了高精度、稳定、可靠的同名点,利用直接解加严密解法结合高精度同名点计算影像间的定向参数。实验证明本文提出的多重约束匹配算法可提高像对间定向的精度及稳定性。
For the problems of poor accuracy and instability of the orientation parameters of non-metric camera pairs,a robust algorithm of image orientation based on SIFT matching under multiple constraints is proposed. First,the scale-invariant SIFT algorithm is used to extract the common points between the images. Then,the left-right symmetry consistency checking,random sampling,and affine transformation are used to perform step-wise constraint matching,which provides high-precision,stable,and reliable common points for orientation. At last,the direct and strict solution method based on the high-precision common points are used to calculate images orientation parameters. Experiments show that the multiconstraint matching algorithm proposed in this paper can improve the accuracy and stability of image orientation.
作者
李新锋
张旭
Li Xinfeng;Zhang Xu(Shandong Zhengvuan Digital City Construction Co.Ltd.,Yantai 264670,China)
出处
《工程勘察》
2020年第8期72-78,共7页
Geotechnical Investigation & Surveying
关键词
SIFT匹配
同名点
仿射变换
定向参数
SIFT matching
common points
affine transformation
orientation parameters
