Spatio-temporal analysis of multiview video

In this thesis, different approaches and algorithms are presented which aim at enabling and facilitating the analysis of multiview video. The temporal and spatial aspects of video are treated separately. The specific contributions are a novel, fully automatic video synchronization algorithm that is highly flexible and very robust against disturbing influences. Furthermore, the thesis introduces a universal framework to evaluate the precision of multiscale image features, as well as a new class of visual markers and two novel and highly efficient multiscale feature detectors.