Wall Estimation from Stereo Vision in Urban Street Canyons

Tobias Schwarze, Martin Lauer

2013

Abstract

Geometric context has been recognised as important high-level knowledge towards the goal of scene understanding. In this work we present two approaches to estimate the local geometric structure of urban street canyons captured from a head-mounted stereo camera. A dense disparity estimation is the only input for both approaches. First, we show how the left and right building facade can be obtained by planar segmentation based on random sampling. In a second approach we transform the disparity into an elevation map from which we extract the main building orientation. We evaluate both approaches on a set of challenging inner city scenes and demonstrate how visual odometry can be incorporated to keep track of the estimated geometry.

Paper Citation

in Harvard Style

Schwarze T. and Lauer M. (2013). Wall Estimation from Stereo Vision in Urban Street Canyons . In Proceedings of the 10th International Conference on Informatics in Control, Automation and Robotics - Volume 2: ICINCO, ISBN 978-989-8565-71-6, pages 83-90. DOI: 10.5220/0004484600830090

in Bibtex Style

@conference{icinco13,
author={Tobias Schwarze and Martin Lauer},
title={Wall Estimation from Stereo Vision in Urban Street Canyons},
booktitle={Proceedings of the 10th International Conference on Informatics in Control, Automation and Robotics - Volume 2: ICINCO,},
year={2013},
pages={83-90},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0004484600830090},
isbn={978-989-8565-71-6},
}

in EndNote Style

TY - CONF
JO - Proceedings of the 10th International Conference on Informatics in Control, Automation and Robotics - Volume 2: ICINCO,
TI - Wall Estimation from Stereo Vision in Urban Street Canyons
SN - 978-989-8565-71-6
AU - Schwarze T.
AU - Lauer M.
PY - 2013
SP - 83
EP - 90
DO - 10.5220/0004484600830090