Autocalibration of multiprojector CAVE-like immersive environments

Area of Science:

• Computer Vision
• Virtual Reality
• Display Technology

Background:

• Immersive display systems like CAVEs (Computer-Aided Virtual Environments) offer high levels of immersion but often rely on complex calibration methods.
• Existing calibration techniques for CAVE-like displays treat complex surfaces piecewise, limiting projector placement and overlap, especially around corners.
• Automatic calibration is crucial for simplifying the setup and enhancing the usability of advanced immersive environments.

Purpose of the Study:

• To present a novel method for geometric autocalibration of multiple projectors on CAVE-like swept surfaces.
• To enable markerless projector registration using a single uncalibrated camera, accommodating nonlinear distortions.
• To facilitate arbitrary projector positioning and achieve viewpoint-independent registration for various virtual reality applications.

Main Methods:

• Geometric autocalibration of projectors on swept surfaces (truncated domes, 4/5-wall CAVEs) using a single uncalibrated camera.
• Handling of nonlinear projector distortions and registration across multiple camera views when the entire display is not visible.
• Recovery of the 3D display surface geometry for accurate projector mapping.

Main Results:

• Successful markerless, automatic geometric registration of multiple projectors on complex CAVE-like surfaces.
• The method supports arbitrary projector placement, including overlapping corners in 5-wall CAVEs, without manual markers.
• Achieved registration adaptable for both single-user head-tracked systems and multi-user collaborative explorations.

Conclusions:

• This autocalibration technique simplifies the deployment of advanced immersive displays, making swept surfaces more accessible.
• Enables the creation of more seamless and immersive virtual reality experiences by allowing flexible projector configurations.
• Opens new possibilities for utilizing complex display geometries in VR without compromising calibration simplicity.