Ask Your Question

Can a paper printed chessboard affect camera calibration?

asked 2016-08-31 02:57:19 -0600

Hilman gravatar image

updated 2016-08-31 06:31:50 -0600

In calibrating my camera, I used a paper printed chessboard. I guess because of the nature of the paper, it is not "perfect" since there will be some crumples/wrinkles. In 15 images that I have used, below are 3 of the examples of the original image, and its undistorted version.

Original image 1 Undistorted image 1

Original image 2 Undistorted image 2

Original image 3 Undistorted image 3

As you can see, there are some wrinkles at the paper (especially at the edge). Can this somehow greatly affects the camera matrix and distortion coefficients that will be calculated, or I am just being overly paranoid here?


edit retag flag offensive close merge delete



" I am just being overly paranoid here?" - not so, imho.

yes, creaks and wrinkles will affect your calibration, so try to glue your paper sheet onto something straight.

(also, it should have a ~1square white border on the top/left side)

berak gravatar imageberak ( 2016-08-31 03:01:35 -0600 )edit

For more details refer this nice answer

Balaji R gravatar imageBalaji R ( 2016-08-31 06:43:00 -0600 )edit

You may also keep in mind that the printer itself will also affect the square size. Meaning that when your square should be 4x4mm it may be sometimes 3.9x4mm sometimes 3.85x3.9mm usw. If it should be accuracte i suggest the solution FooBar mentioned.

Ice_T02 gravatar imageIce_T02 ( 2016-09-01 04:42:03 -0600 )edit

imho it needs more white "space" at those borders.

berak gravatar imageberak ( 2016-09-03 21:29:19 -0600 )edit

He's suggesting more white space as a 'buffer' around the edges as it can make the corner-finding algorithm more robust. If there's not enough white at the edges, the corner-finding may fail sometimes. But it's mainly a robustness issue; if your corners are being detected at all (seems like they are), then probably it's not affecting the accuracy as such. It's just that in future, it's better to have more white to guarantee that the corners will be detected in most conditions.

AJW gravatar imageAJW ( 2016-09-04 05:06:43 -0600 )edit

thanks, @AJW, exactly.

berak gravatar imageberak ( 2016-09-04 05:16:08 -0600 )edit

2 answers

Sort by ยป oldest newest most voted

answered 2016-08-31 10:10:12 -0600

I prefer to create my calibration targets from aluminum composites: They are planar enough, can be rather large and are extremely cheap.

edit flag offensive delete link more

answered 2016-09-04 05:25:20 -0600

AJW gravatar image

You aren't being paranoid; if those wrinkles affect the planarity of the corner points, then it may very well degrade the calibration result. But you'll never make the absolute perfect calibration target, so you really need to be thinking about how much it might be affecting the results. It's best if you start with your application requirements, and then work backwards to figure out how much accuracy you really need.

For example, the undistorted images you show look reasonable; clearly you're getting at least some decent distortion coefficients. If you only need roughly undistorted images, then you're already fine. If you need more accuracy than that, then decide how much, otherwise you may find you go down a rabbit hole chasing accuracy you don't need.

There's no reason to rule out paper a priori - you can certainly get good calibrations with paper targets. But you may want to take more care in mounting the paper. And in some environments the paper might, for example, absorb moisture and wrinkle. Again, depends on your application and usage environment.

edit flag offensive delete link more

Question Tools



Asked: 2016-08-31 02:57:19 -0600

Seen: 13,698 times

Last updated: Sep 04 '16