Indoor-Outdoor 3D Reconstruction Alignment
IEEE European Conference on Computer Vision 2016
Abstract
Structure-from-Motion can achieve accurate reconstructions of urban scenes. However, reconstructing the inside and the outside of a building into a single model is very challenging due to the lack of visual overlap and the change of lighting conditions between the two scenes. We propose a solution to align disconnected indoor and outdoor models of the same building into a single 3D model. Our approach leverages semantic information, specifically window detections, in multiple scenes to obtain candidate matches from which an alignment hypothesis can be computed. To determine the best alignment, we propose a novel cost function that takes both the number of window matches and the intersection of the aligned models into account. We evaluate our solution on multiple challenging datasets.
Publication
Indoor-Outdoor 3D Reconstruction Alignment,
A. Cohen,
J. L. Schönberger,
P. Speciale,
T. Sattler,
J-M. Frahm,
M. Pollefeys.
European Conference on Computer Vision (ECCV) 2016
[PDF] [Video]
[bibtex]
@InProceedings{CohenSchoenb_2016_ECCV,
author = {Cohen, Andrea and Sch\"{o}nberger, Johannes Lutz and Speciale, Pablo and Sattler, Torsten and Frahm, Jan-Michael and Pollefeys, Marc},
title = {Indoor-Outdoor 3D Reconstruction Alignment},
journal = {European Conference on Computer Vision (ECCV)},
location = {Amsterdam, Netherlands},
month = {October},
year = {2016},
}
Datasets
We provide both the images and the sparse reconstructions used in our paper. The reconstructions format is as described here. Note that the datasets House-1 and House-2 are not available due to privacy issues. The remaining datasets can be downloaded from here under the following names:
Theatre : is called gerrard-hall.
Hall : is called graham-hall.
Chapel : is called person-hall.
| 
