In this presentation I will give an overview on the research activities of the computer graphics
group at the University of Tübingen on depth reconstruction from videos including structure from
motion, SLAM and fast nearest-neighbor search. We propose real-time structure from motion for high
spatial and temporal resolution video streams with thousands of frames. The key technical
contributions are a robust selection of confident frames, a novel windowed bundle adjustment,
robust frame-to-structure verification for globally consistent reconstructions with multi-loop
closing, and the utilization of an efficient global linear camera pose estimation that links both
consecutive and distant bundle adjustment windows. In order to allow for efficient loop closing we
developed a fast high-dimensional nearest neighbor search with significant speed-ups compared to
existing libraries. The last part will cover real-time SLAM techniques for robust environment
reconstruction and tracking in stereo car videos.
Hendrik P. A. Lensch holds the chair for computer graphics at Tübingen University and is head of
the computer science department. He received his diploma in computers science from the University
of Erlangen in 1999. He worked as a research associate at the computer graphics group at the
Max-Planck-Institut für Informatik in Saarbrücken, Germany, and received his PhD from Saarland
University in 2003. Hendrik Lensch spent two years (2004-2006) as a visiting assistant professor at
Stanford University, USA, followed by a stay at the MPI Informatik as the head of an independent
research group. From 2009 to 2011 he has been a full professor at the Institute for Media
Informatics at Ulm University, Germany. In his career, he received the Eurographics Young
Researcher Award 2005, was awarded an Emmy-Noether-Fellowship by the German Research Foundation
(DFG) in 2007 and received an NVIDIA Professor Partnership Award in 2010. His research interests
include 3D appearance acquisition, computational photography, global illumination and image-based
rendering, and massively parallel programming.
for more details.