Quasi-Dense Matching between Perspective and Omnidirectional Images

In this paper, we propose a match propagation algorithm between perspective and omnidirectional images of same scene consisting of planes, which does not require a rectification for the omnidirectional image. First, a linear transformation is introduced to identify the area containing the corresponding point candidates. Then, a geometric invariant is computed as a constraint for quasi-dense matching. Finally, combining the computed geometric invariant with a best-first strategy of Lhuillier and Quan (2002), the quasi-dense point correspondences are calculated. The experiments with real data show that the algorithm of this paper has good performance.