Setup-Independent Full Projector Compensation

Area of Science:

• Computer Vision
• Computer Graphics
• Image Processing

Background:

• Projector compensation corrects distortions on nonplanar surfaces.
• Existing methods are setup-dependent, requiring retraining for new configurations.
• Limited datasets and rigid models hinder generalization.

Purpose of the Study:

• Introduce SIComp, the first setup-independent projector compensation framework.
• Enable generalization to unseen projector-camera setups without fine-tuning.
• Address limitations of existing, setup-specific compensation methods.

Main Methods:

• Constructed a large-scale dataset with 277 real-world projector-camera setups.
• Developed a co-adaptive framework decoupling geometry and photometry.
• Utilized an optical flow module for online geometric correction and a photometric network for compensation.
• Integrated intensity-varying surface priors for robustness under varying illumination.

Main Results:

• SIComp achieves high-quality compensation across diverse, unseen setups.
• Demonstrated superior generalization ability compared to existing methods.
• Established the first generalizable solution for full projector compensation.

Conclusions:

• SIComp offers a robust and generalizable solution for projector compensation.
• The framework overcomes the setup-dependency limitations of prior work.
• Enables reliable image projection correction in dynamic environments.