Calibration of smooth camera models.

Area of Science:

• Computer Vision
• Computational Imaging
• Geometric Modeling

Background:

• Current generic imaging models are discrete, mapping image pixels to 3D lines.
• Existing models present challenges in calibration complexity and resolution.
• A need exists for more efficient and accurate generic camera representations.

Purpose of the Study:

• To present a modified generic camera model using radial basis functions (RBFs).
• To develop a simplified and more accurate camera calibration procedure.
• To enhance resolution and achieve a more compact representation of imaging models.

Main Methods:

• Modified the general imaging model using radial basis functions (RBFs) for interpolation.
• Developed a calibration procedure requiring only 3D point-to-pixel matching.
• Applied normalization to image and 3D point coordinates to improve calibration accuracy.

Main Results:

• The proposed continuous model increases resolution and offers a compact representation.
• The calibration procedure is significantly simplified, requiring fewer calibrated points.
• Both synthetic and real-world data demonstrated accurate and applicable calibration results.

Conclusions:

• The RBF-based generic camera model offers advantages in resolution and representation compactness.
• The simplified calibration procedure reduces complexity and enhances applicability.
• The method provides accurate calibration, suitable for diverse computer vision tasks.