Extrinsic Calibration of Camera and Motion Capture Systems

Abstract

When working with camera sensors, especially in research and development efforts, it is often important to be able to locate the camera optical coordinate frame with reference to a global coordinate frame. In this work, we use a motion capture system and fiducial markers to estimate the translational and rotational offsets between the camera frame relative to a tracked motion capture (MoCap) marker frame. The solution provides tools critical for creation, validation, experimentation and deployment of imaging systems with far reaching applications in vision systems and mobile robotics. The proposed calibration protocol uses three pose estimates: thepose of camera w.r.t fiducial markers, the pose of fiducial marker w.r.t MoCap defined global frame and the pose of camera marker w.r.t. MoCap defined global frame to produce a real-time stream of the calibration parameters. The Robot Operating System (ROS) serves to interpolate sensed pose data to create time-synchronous pose estimates from which the extrinsic parameters can be directly calculated. We provide detailed analysis of our results and demonstrate the consistency and robustness of this approach. The link to the open-source code and detailed documentation is also provided <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">11</sup> GitHub Repository.