Abstract:

In the field of computer vision and close range photogrammetry, it is critical to accurately calibrate the intrinsic parameters and the fixed relative orientation relationship between the two cameras of the stereo-camera system used for high precision 3D reconstruction. In this paper, six fixed and independent relative orientation constraints between the two cameras are derived firstly, and then self-calibration bundle adjustment with restrictive constraints is employed to calibrate the stereo-camera system. Finally, synthetic and real data are both employed to verify the performance of the calibration method proposed in this paper. The experiment result reveals that the proposed relative orientation constraints can effectively increase the redundant observations of the calibration adjustment system, and this can improve the accuracy and stability of the calibration. This method can also be extended to other stereo-camera systems which consist of more than two cameras.

Key words: camera calibration, stereo-camera, relative orientation relationship, 3D reconstruction, bundle adjustment