Ask Your Question
0

Data for camera calibration input

asked 2017-02-28 20:15:40 -0600

I am new to the concept of camera calibration and trying to understand the theoretical concept. My question is why should I move my targets at different orientation in space. What if, I just fix my calibration target and just move the camera in X and Y direction without any tilt across the entire image space.

I tried this approach and got an average reprojection error of 0.5. I was wondering if this result is valid or is it really necessary to tilt the camera and take more data?

I am really sorry if its a lame question. This is the first time I am doing camera calibration

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
0

answered 2017-02-28 21:09:11 -0600

Tetragramm gravatar image

So long as the calibration points thoroughly cover the image space you're ok.

Its typically easier for someone to hold up a chessboard at different locations, but if you have the setup to do as you say, it's fine.

edit flag offensive delete link more

Comments

Hi thank you for your answer. But is there any paper or argument that validates this approach? because from whatever I have read so far, the general rule of calibration is to have translation and rotation of either the calibration target or the camera while taking pictures.

anand.jagan gravatar imageanand.jagan ( 2017-03-01 11:41:32 -0600 )edit

Huh. Obviously I need to take a deeper look at the math. There's a singularity on the focal length. You can undistort pretty well, based on the guess of the center pixel is the principal point, but the focal length is unknowable. I verified this with some test code.

Reference: https://www.vision.caltech.edu/bouguetj/calib_doc/papers/sturm99.pdf

I ran a test program using projectPoints. It doesn't seem to like any of the translation only tests I ran. If shifting in X and Y, distortion seems to be ok, if pretty rough. Depth only made an ok focal length, but the distortion was not altered at all.

Tetragramm gravatar imageTetragramm ( 2017-03-01 19:16:16 -0600 )edit

Question Tools

1 follower

Stats

Asked: 2017-02-28 20:12:09 -0600

Seen: 91 times

Last updated: Feb 28 '17