Homography estimation based on common self-polar triangles of sphere images
Yue Zhao, Xuechun Wang, Xiaojuan Yu · Journal of Modern Optics · 2020
In the unit viewing sphere imaging model, the projection of a space sphere includes the sphere image and its antipodal sphere image. As the pair of sphere images are separated, they have only one common self-polar triangle. In this study, a method for homography estimation based on the projection of the space sphere is proposed, which involves using a paracatadioptric camera. The proposed method requires two images of the space sphere from different positions to obtain a homography matrix. This homography matrix has eight degrees of freedom, and a line correspondence pair provides two constraints. Hence, using four pairs of line correspondences, the homography matrix can be solved, and the external parameters can be obtained. The effectiveness of the proposed method is verified through simulation and experiments using real images.