I have a small doubt, that do we require multiple images of the calibration target pattern to precisely calibrate the camera. Because, matlab doesn’t allow me to just use a single image of the calibration target for performing the calibration routine. Why is this the case that we require more than two images with different orientations. Also, suppose we just want to calibrate the camera for a 2d field, does it make sense to just a single image?
Define “precise”. Generally, the number of images with the target affects the accuracy of the calibration.
1 Like
So, more the images better the calibration right?
But do the calibration target images need to be in different orientations ?
I’m not sure what exactly you’re asking about. If you’re using Matlab’s Camera Calibration Toolbox then I suggest you read their (excellent) documentation.
Otherwise, it’s better to simply start a new topic for those kind of questions (this specific topic is 2 years old), or better yet, ask on a Matlab discourse 
1 Like
I believe a photometric method will provide a higher accuracy calibration than feature based methods (such as the corner detection that tools such as the MATLAB camera calibration toolbox use). This is because photometric methods minimize errors in the domain of the true measurement noise, which is pixel intensity, not geometric error.
I am seeking collaborators to provide Julia with a state of the art photometric optimization package. See here for more details.
1 Like