Extrinsic calibration of a non-overlapping camera network based on close-range photogrammetry

Area of Science:

• Computer Vision
• Photogrammetry
• Robotics

Background:

• Camera network calibration is crucial for 3D reconstruction and spatial awareness.
• Existing methods often require camera movement or specialized calibration targets.
• Non-overlapping camera networks present unique calibration challenges.

Purpose of the Study:

• To develop an extrinsic calibration method for non-overlapping camera networks.
• To enable calibration without camera movement or dedicated targets.
• To achieve accurate 3D reconstruction and pose estimation.

Main Methods:

• Utilized close-range photogrammetry with arbitrarily distributed encoded targets.
• Employed a three-step calibration process: 3D target reconstruction, intrinsic calibration, and extrinsic parameter estimation.
• Leveraged a hand-held digital camera for initial target localization.

Main Results:

• Achieved a relative error of less than 0.003% for 3D reconstruction.
• Demonstrated relative errors below 0.066% for rotation and translation.
• Reported a re-projection error of only 0.09 pixels, indicating high accuracy.

Conclusions:

• The proposed method offers an efficient and accurate solution for extrinsic calibration of non-overlapping camera networks.
• It eliminates the need for camera movement and calibration targets, simplifying the process.
• The validated results confirm the method's effectiveness for various applications requiring precise camera pose estimation.