Accurate feature detection for out-of-focus camera calibration

Area of Science:

• Computer Vision
• Photogrammetry
• Optical Metrology

Background:

• Conventional camera calibration requires well-focused images for accurate feature detection.
• This limitation poses practical challenges in image acquisition for both long- and short-distance photogrammetry.
• Existing methods are sensitive to image quality and defocusing.

Purpose of the Study:

• To develop a robust camera calibration method insensitive to image defocusing.
• To enable accurate feature point extraction from blurred or unfocused calibration patterns.
• To enhance the practicality and convenience of image acquisition in photogrammetry.

Main Methods:

• Utilized three active phase-shift circular grating (PCG) arrays as calibration patterns.
• Employed ellipse fitting of 2π-phase points for accurate extraction of PCG centers as feature points.
• Simulated image blurring using Gaussian filters with varying standard deviations and captured images with different defocusing degrees.

Main Results:

• The proposed method accurately extracts feature points from substantially blurred pattern images.
• The accuracy and reliability of the method were validated across various degrees of blurring and defocusing.
• The method demonstrated insensitivity to image defocusing, pattern period, number of gratings, and noise.

Conclusions:

• The phase-shift circular grating (PCG) based camera calibration method is accurate and reliable.
• This approach overcomes the limitation of requiring well-focused images, enhancing practical usability.
• The method offers a robust solution for camera calibration in challenging imaging conditions.