Epipolar error for camera calibration.

edit

Hi,

As seen from the tutorial on stereo camera calibration here, after calibration, it seems that ROS reports the epipolar error as a good measure of the camera calibration.

Are there any research papers that support the fact that the epipolar error is enough to qualitatively measure camera calibration ? What about measures like average (over all pairs of images) re-projection error (While using OpenCV directly I came across this here )

Thanks!