The reconstruction of 3D scene models from arbitrary video sequences is a central task of computer vision. Applications range from the autonomous navigation of robots to the acquisition of 3D models for the entertainment industry. Typical approaches identify and track landmarks in the images as a first step and then recover the geometry in a separate, second step. This book instead focuses on a combination of the two steps. By embedding the novel, combined method in a model-based mathematical framework, superior accuracy and reliability can be shown both theoretically and empirically.

Olaf Kähler, born in 1979, studied computer science at the Friedrich-Alexander-Universität Erlangen-Nürnberg. From 2005 to 2009 he was then working at the Chair for Computer Computer Vision of the Friedrich-Schiller-Universität Jena and in that course focused on the 3D reconstruction and structure-from-motion problems.