Microsoft and the University of Washington Team Up to Create an Immersive Online Photo-Sharing Experience

WebKnowHow
Wednesday, August 2, 2006; 09:22 AM

At SIGGRAPH 2006 this week, researchers from Microsoft Corp. and the University of Washington will present a new photo-browsing system that enables people to combine their photos with thousands of others collected on the Internet to present a detailed 3-D model which gives viewers the sensation of smoothly gliding around the scene from every angle.

"What if your photo collection was an entry point into the world, like a wormhole that you could jump through and explore… "

“Microsoft Research has been successfully transferring technology into key Microsoft products for several years. Photosynth is the first of many upcoming examples of Live Labs’ ability to take basic research from Microsoft Research and the academic community and rapidly create prototypes of new online services,” said Dr. Gary William Flake, noted industry technologist and Microsoft technical fellow.

Noah Snavely, a graduate student in the department of computer science and engineering at the University of Washington, and associate professor Steven Seitz collaborated with Microsoft researcher Richard Szeliski to develop the technology that was the genesis of Photosynth. Their work combined both new and established techniques in computer vision; image-based modeling and rendering; image browsing, retrieval and annotation; and key-point detection and matching to present unstructured collections of photos in a 3-D perspective. As people navigate around the model with their mouse and click on specific photos, the system smoothly transitions between images to create an evocative sense of movement.

“It’s like a hybrid of a slide show and a gaming experience that lets the viewer zoom in to see greater detail or zoom out for a more expansive view,” said Szeliski, manager of the Interactive Visual Media Group within Microsoft Research. “This is a revolutionary way for people to interact with photos in a 3-D context that more closely resembles the place where the images were captured.”

Demos, screen shots, the research paper and other information on Photo Tourism can be found on the University of Washington (http://phototour.cs.washington.edu) and Microsoft Research (http://www.research.microsoft.com/IVM/PhotoTours)Web sites.

Microsoft Research teams with scientists at scores of universities and other organizations each year as part of the company’s commitment to expanding the boundaries of computer science and related fields. “Working with Microsoft Research and Live Labs on this project has been a rewarding experience for me as well as an invaluable opportunity for my student Noah to pioneer a technology that can impact a huge user base,” Seitz said. “The collaborative environment brought a greater level of expertise, sparking some important ideas and bringing results to the project that we might not have otherwise seen.”

Snavely, Seitz and Szeliski will present their research paper detailing the technology on Aug. 2 at SIGGRAPH 2006, the 33rd annual international conference and exhibition on computer graphics and interactive techniques. The gathering of roughly 25,000 attendees is sponsored by ACM SIGGRAPH, the leading professional society for computer graphics and interactive techniques.

Microsoft Research’s association with SIGGRAPH spans more than a decade. Researchers from Microsoft’s labs in Beijing, China; Cambridge, England; and Redmond, Wash., have contributed 17 of the 86 papers accepted for SIGGRAPH 2006, with 11 of those papers resulting from Microsoft Research collaborations with various universities.

Among these projects is a technology system for drag-and-drop pasting of an object or region from one source image — such as a detail of a surfer or a sand pyramid within a photo — onto another image in such a way that they blend seamlessly. Researchers at the Chinese University of Hong Kong, Microsoft Research Asia and the Hong Kong University of Science and Technology co-developed this user-friendly image-composition tool.

Another paper authored by four researchers from Microsoft Research Cambridge describes AutoCollage, an automatic procedure for efficiently constructing a visually appealing photo layout from a set of images. Key features of the AutoCollage technology include its ability to combine more than 50 images; transparently blend the seams between images; and recognize characteristics of different objects, such as sky and faces, to position and depict them appropriately.