Exploring Community Photo Collections

People

Abstract

How can we take advantage of the huge images databases available nowadays in photo-sharing Web sites like Flickr and Google Images to build models of the world surrounding us? Taking that question into consideration, this work is a step towards a framework capable of performing analysis of photographs to build sparse 3D representation of the environment captured in those pictures. Our system integrates many available tools to achieve a robust Structure from Motion solution to gather 3D information from photos.

This project was developed under supervision of teacher Luiz Velho as a partial requirement for the discipline “Fundamentals and Trends in Image Processing” from IMPA, offered in the second semester of 2008.

Download

proposal thumbnail

Proposal

Document describing the goals in the project. Submitted to Luiz Velho in the beginning of the course. (.pdf file, 30 KB)






presentation thumbnail

Method

Slides used for a talk on the project at the end of the course. For a detailed description of the general method, please refer to this document. (.pdf file, 3 MB)


videos thumbnail

Videos

Some videos showing the resulting reconstructed scenes.

Files ballet.avi and breakdancers.avi were generated using one frame of the sequences provided by the Interactive Visual Media Group at Microsoft Research.
File et.avi was generated using data provided by the University of Washington and Noah Snavely as part of their SfM code. (.zip file, 7 MB)

References

[1] C.L. Zitnick, S.B. Kang, M. Uyttendaele, S. Winder, and R. Szeliski. "High-quality video view interpolation using a layered representation." ACM SIGGRAPH and ACM Trans. on Graphics, Los Angeles, CA, Aug. 2004, pp. 600-608.

[2] Richard Hartley, Andrew Zisserman. "Multiple View Geometry in Computer Vision, second edition." Cambridge University Press, ISBN: 0521540518, 2004.

[3] Michael Goesele, Noah Snavely, Brian Curless, Hugues Hoppe, Steven M. Seitz. "Multi-View Stereo for Community Photo Collections." Proceedings of ICCV 2007, Rio de Janeiro, Brasil, October 14-20, 2007.

[4] Noah Snavely, Rahul Garg, Steven M. Seitz, and Richard Szeliski. "Finding Paths through the World's Photos." ACM Transactions on Graphics (SIGGRAPH Proceedings), 27(3), 2008, 11-21.

[5] Noah Snavely, Steven M. Seitz, Richard Szeliski. "Photo tourism: Exploring photo collections in 3D." ACM Transactions on Graphics (SIGGRAPH Proceedings), 25(3), 2006, 835-846.

[6] http://grail.cs.washington.edu/projects/cpc/

[7] http://www.cs.unc.edu/~ccwu/siftgpu/

[8] http://www.cs.umd.edu/~mount/ANN/

Acknowledgments

Special thanks to University of Washington and Noah Snavely for making their SfM code available for the scientific community, and to Changchang Wu for his efforts in building SiftGPU.