Multiple-view geometry for image-based modeling
Jana Košecká, Yi Ma, Stefano Soatto, René Vidal · 2004
This course presents the state of the art in multiple-view geometry, including methods and algorithms for reconstructing 3-D geometric models of scenes from video or photographs. This course is based on a novel approach to multiple-view geometry that only requires linear algebra, as opposed to more involved projective and algebraic geometry that most current methods employ. This new approach aims to make image-based modeling techniques accessible to a larger audience compared to existing ones. The presentations will be based on a recent Springer-Verlag textbook on 3D vision (coauthored by the course organizers). Matlab code for the algorithms will be made available at the book website.