Ask Your Question

Callibrating a bunch of cameras - Can I use a linear rail robot?

asked 2017-01-06 11:06:24 -0500

TAXfromDK gravatar image

updated 2017-01-06 11:07:45 -0500

Hi Guys,

I need to calibrate a bunch of cameras, and I was wondering if the opencv camera calibration could work if I build a simple robot with a screen and a linear rail to move the camera. Something like this: image description

I want to use a robot so I can reproduce the same calibration action across multiple cameras.

My question is if it will work only to move the camera linearly relative to the checkerboard or if rotation is required as well?

Kind regards


edit retag flag offensive close merge delete


i cannot answer,- but definitely - +1 for physical effort.

berak gravatar imageberak ( 2017-01-06 13:50:13 -0500 )edit

2 answers

Sort by ยป oldest newest most voted

answered 2017-01-07 12:16:37 -0500

Eduardo gravatar image

In my opinion, using a robot to calibrate is a good approach as you should get:

  • repeatable and consistent results,
  • accurate calibration once you find a good predefined set of poses.

As you should be able to generate only a translational motion along the rail, you will have in my opinion to generate multiple translation / orientations (rotations) for the pattern. Or move the rail in translation and in rotation. The ideal case I think would be to cover the space in an hemisphere above the calibration pattern.

The best would be to use a 6 dof robot to move the calibration pattern like in this video.

Almost the same topic: Industrial Calibration Library Update and Presentation.

edit flag offensive delete link more


You sure? He's translating into (or out of) the screen, so the points are going to basically move radially outward from the center as he moves closer. That should pretty thoroughly cover the area even without generating multiple sets.

Tetragramm gravatar imageTetragramm ( 2017-01-07 12:27:13 -0500 )edit

I was always told the following advices for camera calibration (with no guarantee that they are correct!):

  • calibrate at the distance you will use for the real case
  • the calibration pattern must occupy most of the part of the image
  • use multiple camera orientations / positions / translations to cover an hemisphere above the pattern
  • good reprojection error should be less than 0.5 px

Here are some example images for camera calibration.

Eduardo gravatar imageEduardo ( 2017-01-08 08:14:16 -0500 )edit

Thank you for your replies.

I will attempt calibrating and report back when I know more. Kind regards


TAXfromDK gravatar imageTAXfromDK ( 2017-01-09 15:28:36 -0500 )edit

answered 2017-01-06 20:25:24 -0500

Tetragramm gravatar image

Yes, I believe it should work. I haven't tried it, but you will get your points radially expanding outward from the center and cover most of the parts of the camera FPA. That should work.

You do need to be careful that you can either see the whole target board at the same time, or that you adjust the number of rows and columns to match what the camera is seeing or you won't be able to detect the corners properly.

Use the drawChessboardCorners like in the tutorial to make sure it looks right.

edit flag offensive delete link more


Thank you for your replies.

I will attempt calibrating and report back when I know more. Kind regards


TAXfromDK gravatar imageTAXfromDK ( 2017-01-09 15:28:48 -0500 )edit

Question Tools

1 follower


Asked: 2017-01-06 11:06:24 -0500

Seen: 176 times

Last updated: Jan 07 '17