ECCV 2004 Tutorial T5

Date: Monday May 10, 2004 morning   Venue: Zofin Palace

Camera-Based Techniques for Building Large-Area Multi-Projector Displays

Instructors: Aditi Majumder, Michael S. Brown

Duration: 4 hours


In the computer graphics and scientific visualization community, large-scale, multi-projector displays serve as the premiere way to visualize and interactive with CG imagery. Multi-projector display environments, however, have been traditionally very difficult and costly to build. This is due to the fact that most projector-based displays require precise manual positioning of the projectors as well as precise display surface construction. Recently, camera-based computer vision techniques have been introduced that can automatically register the geometry of multiple projectors to produce seamless imagery with much less rigid designs constraints. These techniques allow for very flexible display environments and future work has the potential to change how we interactive with computer-generated imagery.

The goal of this tutorial is to introduce to vision researchers an application that has significant research and commercial potential; and to impart a working knowledge of the current state-of-the-art in projector-based displays and the many remaining short-term and learn-term challenges.

While our main audience has traditionally targeted graphics researchers, with previous tutorials offered at SIGGRAPH and EuroGraphics, in this tutorial we focus our attention to the technical details and problems pertaining to projector-based graphics that will be of interest to computer vision researchers. In particular, we focus on two fundamental challenges: (1) geometric registration of projected imagery with the physical world, and (2) photometric registration for producing perceptually consistent imagery.

Target Audience:

The target and audience interested in detailed knowledge of the challenges, current solutions, and open problems related to computer vision for building large-scale multi-projector displays. It can also serve as a good introductory tutorial for any one interested in projection based tiled displays in general.

This tutorial will be self contained and does not assume prior knowledge on large area displays. Familiarity and prior exposure to large scale visualization will be helpful. A basic understanding of some mathematics, such as linear algebra is required.

Topics to be covered:

On the instructors:

Michael S. Brown is an Assistant Professor of Computer Science at the Hong Kong University of Science and Technology. His research is focused on camera-based geometric registration of multi-projector displays and the integration of these approaches into PC-based rendering architectures.

Aditi Majumder is an Assistant Professor of Computer Science at the University of California at Irvine and a Research Associate at Mathematics and Computer Science Division of Argonne National Laboratory. Her research and PhD thesis is focused on color seamlessness of projection based displays and human computer interaction with large tiled displays.