is the calibration parameters differs when the distance between camera and check board pattern varies?

hi everyone, i have a camera set up with 5 camera's placed at 35 degrees apart around the boundary of a semicircle. On trying to calibrate, the parameters are varying whenever i change the position of the cameras i.e, moving all cameras with equal distance away from center of semicircle. Does the calibration of camera depend on depth? actually i have a set of image database taken from those cameras. But now the position of the cameras have changed in the camera set up.So if i calibrate now ,with the changed position of the cameras, will the calibration parameters be applicable to those images or not?

 Accepted Answer

Hi Ram,
Could you please clarify which camera parameters you are talking about? Intrinsic camera parameters (optical center and the focal length) do not depend on camera's position. Neither do the distortion coefficient. On the other hand, camera extrinsics - rotation and translation relative to some world coordinate system - obviously do.
Also, keep in mind that while the actual camera intrinsics and distortion do not depend on camera's position, your estimate of them might. If you calibrate by taking images of a checkerboard 1 meter away vs. 5 meters away, you will get slightly different values.

3 Comments

thanks dima lisin for your answer. actually i have 5 camera setup as i said above. i took images of checkerboard and calibrated them using camera calibrator app in matlab. when i changed the position of camera's(either left or right ,forward or backward) the intrinsic parameter values are varying. i have a database of images which was taken long back with those camera's. now can i apply these intrinsic parameters to those images or not?
Are you calibrating the cameras individually? How did you determine that the instrinsics are changes when you move the camera? Did you re-calibrate?
yes i re calibrated using camera calibrator app in matlab. i have one more doubt i caluculated fundamental matrix using stereo camera calibrator app in matlab and by using pixel correspondences with ransac algorithm but both results are differed.how it will be possible? which one i have to assume as correct?

Sign in to comment.

More Answers (0)

Categories

Find more on MATLAB Support Package for USB Webcams in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!