March 28, 2024, 02:38:00 am

News:

This site is now on a new server!


Multi-View Stereo for Community Photo Collections

Started by Unbeliever, April 05, 2011, 04:25:32 pm

Previous topic - Next topic

Unbeliever

http://grail.cs.washington.edu/projects/mvscpc/

QuoteAbstract
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.
"Some say God is living there [in space]. I was looking around very attentively, but I did not see anyone there. I did not detect either angels or gods....I don't believe in God. I believe in man - his strength, his possibilities, his reason."
Gherman Titov, Soviet cosmonaut, in The Seattle Daily Ti