Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Multi-view stereo for community photo collections

: Goesele, Michael; Snavely, Noah; Curless, Brian; Hoppe, Hugues; Seitz, Steven M.


Institute of Electrical and Electronics Engineers -IEEE-; IEEE Computer Society:
IEEE 11th International Conference on Computer Vision, ICCV 2007 : 14-21 Oct. 2007, Rio de Janeiro, Brazil
Los Alamitos, Calif.: IEEE Computer Society, 2007
ISBN: 978-1-4244-1630-1
ISBN: 978-1-4244-1631-8
ISBN: 1-4244-1630-2
8 pp.
International Conference on Computer Vision (ICCV) <11, 2007, Rio de Janeiro>
Conference Paper
Fraunhofer IGD ()
computer vision; multi-view stereo; Internet; community photo collection

We present a multi-view stereo algorithm that addresses the extreme changes in lighting, scale, clutter, and other effects in large online community photo collections. Our idea is to intelligently choose images to match, both at a per-view and per-pixel level. We show that such adaptive view selection enables robust performance even with dramatic appearance variability. The stereo matching technique takes as input sparse 3D points reconstructed from structure-from-motion methods and iteratively grows surfaces from these points. Optimizing for surface normals within a photoconsistency measure significantly improves the matching results. While the focus of our approach is to estimate high-quality depth maps, we also show examples of merging the resulting depth maps into compelling scene reconstructions. We demonstrate our algorithm on standard multi-view stereo datasets and on casually acquired photo collections of famous scenes gathered from the Internet.