Reconstructing the World's Museums

Jianxiong Xiao and Yasutaka Furukawa

ECCV 2012 Best Student Paper Award  and  Google Research Excellent Papers Award


Photorealistic maps are a useful navigational guide for large indoor environments, such as museums and businesses. However, it is impossible to acquire photographs covering a large indoor environment from aerial viewpoints. This paper presents a 3D reconstruction and visualization system to automatically produce clean and well-regularized texture-mapped 3D models for large indoor scenes, from ground-level photographs and 3D laser points. The key component is a new algorithm called "Inverse CSG" for reconstructing a scene in a Constructive Solid Geometry (CSG) representation consisting of volumetric primitives, which imposes powerful regularization constraints to exploit structural regularities. We also propose several techniques to adjust the 3D model to make it suitable for rendering the 3D maps from aerial viewpoints. The visualization system enables users to easily browse a large scale indoor environment from a bird's-eye view, locate specific room interiors, fly into a place of interest, view immersive ground-level panorama views, and zoom out again, all with seamless 3D transitions. We demonstrate our system on various museums, including the Metropolitan Museum of Art in New York City -- one of the largest art galleries in the world.



Videos (HD available)

Choose different videos →

Interactive Demo


      author = {Jianxiong Xiao and Yasutaka Furukawa},
      title = {Reconstructing the World's Museums},
      booktitle = {Proceedings of the 12th European Conference on Computer Vision},
      series = {ECCV '12},
      year = {2012},

Interactive 3D Exploration and Tour

Supplementary Materials: View Dependent Model Construction and More Resutls


We would like to acknowledge Google Art Project for providing access to the Museums featured. We thank Google Seattle Lightfield team for helpful discussion and support during this work, especially Carlos Hernandez, David Gallup, and Steve Seitz. We also thank Maneesh Agrawala and Andrew Owens for brainstorming and discussion. This work was done when Jianxiong Xiao interned in Google. Jianxiong Xiao is supported by Google U.S./Canada Ph.D. Fellowship in Computer Vision.

Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of Google Inc.