Received: 30 January 2007 / Accepted: 31 October 2007 | Noah Snavely · Steven M. Seitz · Richard Szeliski
The paper "Modeling the World from Internet Photo Collections" by Noah Snavely, Steven M. Seitz, and Richard Szeliski explores the potential of using the vast collection of internet photographs for 3D scene modeling and visualization. The authors present structure-from-motion (SfM) and image-based rendering algorithms that can process hundreds of images downloaded from keyword searches, enabling reconstructions of well-known sites like Notre Dame and the Trevi Fountain. This approach, called *Photo Tourism*, aims to create 3D models of well-photographed sites, cities, and landscapes from internet imagery. The paper discusses key challenges such as registration and robust processing of diverse image conditions, and outlines future research directions, including more efficient correspondence and reconstruction techniques for large datasets. The authors also describe a photo explorer interface that allows users to virtually explore 3D scenes by moving between photographs, with transitions that emphasize spatial relationships. The paper includes extensive results on 11 scenes and a discussion of related work, highlighting the broad impact of this research on fields like 3D visualization, localization, and communication.The paper "Modeling the World from Internet Photo Collections" by Noah Snavely, Steven M. Seitz, and Richard Szeliski explores the potential of using the vast collection of internet photographs for 3D scene modeling and visualization. The authors present structure-from-motion (SfM) and image-based rendering algorithms that can process hundreds of images downloaded from keyword searches, enabling reconstructions of well-known sites like Notre Dame and the Trevi Fountain. This approach, called *Photo Tourism*, aims to create 3D models of well-photographed sites, cities, and landscapes from internet imagery. The paper discusses key challenges such as registration and robust processing of diverse image conditions, and outlines future research directions, including more efficient correspondence and reconstruction techniques for large datasets. The authors also describe a photo explorer interface that allows users to virtually explore 3D scenes by moving between photographs, with transitions that emphasize spatial relationships. The paper includes extensive results on 11 scenes and a discussion of related work, highlighting the broad impact of this research on fields like 3D visualization, localization, and communication.