The original point of this project concerned the possibility of adjusting the point coordinates given to the calibration routine. Since the points are probably not detected entirely correctly earlier in the process, it is likely that moving them slightly in some educated manner will improve the accuracy of the calibration. For instance, Mark has found from some informal tests that the error in point mapping after calibration by this method is related to the magnitude and direction of the image gradients at detected corner points; moving detected coordinates along the gradient might help fix that problem. Now that the calibrator has been written, it is the intent of this project to test various types of coordinate adjustments for efficacy in improving the accuracy of the resulting calibration.
Although the image gradient is the most obvious tool for adjusting point positions, moving points a random distance in a random direction might also work as long as they are all moved the same way. However, the fact that they can only be moved in two directions and that their movements must somehow all be related are very limiting; there may be no perfect way to adjust them.