TL;DR
This paper introduces a novel method for automatically aligning and correcting geometric distortions in multi-projector display systems, particularly for curved screens. By employing a red-blue coded structured light technique and a new "consistency principle of geometric correction," the authors achieve seamless, large-scale visual experiences crucial for applications like virtual reality.
What Problem Does This Paper Address?
Creating immersive, large-scale visual experiences often involves tiling multiple projectors onto a single display surface. When this surface is curved, two significant challenges arise: geometric distortion (where projected images appear warped) and color inconsistency. While existing methods address geometric correction, there's a lack of theoretical grounding for offline correction. This paper tackles the problem of automatically registering and geometrically correcting multi-projector systems on curved screens, aiming to produce a seamless and distortion-free image for the viewer, regardless of the screen's shape.
Methods
The core of the paper lies in its novel approach to geometric registration. The authors first theoretically demonstrate the "consistency principle of geometric correction," which posits that geometric error is solely dependent on the system's structural parameters, not the virtual scene content. This allows for offline calibration.
Their proposed method utilizes a red-blue gray code (RBGC) structured light technique. This involves projecting sequences of red and blue patterns onto the curved screen. Two pairs of binocular cameras capture these patterns, allowing for:
- 3D Reconstruction of the Curved Screen: By analyzing the captured structured light images, the precise geometry of the curved screen is reconstructed in 3D.
- Feature Point Identification: The RBGC patterns are designed to be robust against ambient light, a common issue with simpler black-white gray codes. The paper details an edge detection method for RBGC that uses divergence to accurately identify feature points where horizontal and vertical stripes intersect.
- Geometric Homography Calculation: Using the reconstructed screen geometry and the captured feature points, a geometric homography is calculated between each projector and the curved screen. This homography forms a "pre-warp template."
- Seamless Display Generation: Projecting images using these pre-warp templates effectively corrects for geometric distortions, resulting in a seamless and aligned image across all projectors on the curved screen.
Datasets
Not discussed in the paper. The authors describe experimental setups and results based on their own captured data from projected structured light images.
Key Insights and Ideas
- Consistency Principle of Geometric Correction: This theoretical contribution is significant. It establishes that geometric correction is an intrinsic property of the display system's structure and can be performed offline, independent of the dynamic content being displayed. This simplifies calibration and allows for pre-computation.
- Red-Blue Gray Code (RBGC) Structured Light: The RBGC approach offers improved robustness against ambient light compared to traditional black-white gray codes. This is crucial for real-world applications where perfect lighting conditions are rare.
- Two-Dimensional Geometric Correction on Curved Surfaces: The paper moves beyond one-dimensional (e.g., horizontal-only) correction for curved screens, proposing a method that handles both horizontal and vertical distortions simultaneously.
Key Results
- Improved Accuracy: Experimental results comparing RBGC with black-white gray code (BWGC) for screen center measurement show that RBGC yields results closer to the physical coordinates, with a smaller distance distribution (0–0.005 m for RBGC vs. 0–0.01 m for BWGC).
- Seamless Display Achieved: A six-projector system on a 1/4 spherical screen demonstrated successful geometric registration, transforming a distorted, misaligned display into a seamless, large-scale picture.
- Offline Calibration Enabled: The theoretical consistency principle allows for the geometric correction process to be performed offline, determining structural parameters without needing real-time scene data.
Limitations and Future Work
- Ambient Light Sensitivity: While RBGC is more robust than BWGC, the paper acknowledges that ambient light still affects accuracy. The proposed edge detection method aims to mitigate this, but further improvements might be needed for highly challenging environments.
- Computational Complexity: While not explicitly detailed as a limitation, the process of 3D reconstruction and homography calculation can be computationally intensive. Future work could focus on optimizing these algorithms for faster real-time applications.
- Color Consistency: The paper primarily focuses on geometric registration. While it mentions color inconsistency as a problem, the proposed method's effectiveness in addressing color blending is not detailed. Future work could integrate color calibration with geometric correction.
Summary updated