Page 1 of 1

Intrinsic parameters and camera orientation

Posted: Tue Nov 12, 2013 4:32 pm
by JeDi
Hi,

We are doing research about different camera calibration techniques, and we are investigating the results we get from the calibration using the Motive::Tracker OptiWand calibration. We are using V120:SLIM cameras with 58 degree FOV lenses. We use active markers in our setup; we converted the calibration wand and ground plane to use active LEDs.

I did the following:
1. I did a wanding calibration session in Motive. I saved the calibration file, opened it using the Motive SDK and output the camera positions, orientation matrices and distortion properties
2. I did a wanding session and output the 2D data (precision mode) to try and triangulate using the above parameters

There are a couple of problems/uncertainties I would like to solve.

The first strange thing we noticed is that if we plot the cameras in matlab using the positions and orientations retrieved from the SDK, the cameras seem to be pointing in the opposite direction that they should be. How should the rotation matrix be interpreted? What is the coordinate system? This is the plotted image:
dcdbddeg.png
Camera position and orientation
dcdbddeg.png (5.01 KiB) Viewed 3117 times
Second, the intrinsic properties retrieved using

Code: Select all

GetDistortionModel
are the same for all cameras. It seems they are the default factory settings for the lenses we use. Could it be that the actual parameters calculated in the calibration are not set? Is there another way I could get the calculated intrinsic parameters?

Thanks in advance for any answers. Also, if there is any more available information about the calibration / triangulation process of unlabelled markers that you could share, I would be very thankful!

Greetings,
JeDi

Re: Intrinsic parameters and camera orientation

Posted: Thu Nov 14, 2013 11:00 am
by NaturalPoint-Dustin
Hello,

Thank you for posting this on the forums. I will have one of our developers submit a response for you shortly.

Regards,

Re: Intrinsic parameters and camera orientation

Posted: Thu Nov 14, 2013 11:13 am
by beckdo
Hey Jedi,

The calibration results are left-handed coordinates currently. In your case you should be able to rotate the cameras 180 in order to point them in the correct direction.

Unfortunately there is no way to get the solved intrinsic parameters out of our system. The only thing available is the default intrinsic parameters for the cameras currently.

D

Re: Intrinsic parameters and camera orientation

Posted: Fri Nov 15, 2013 3:40 am
by JeDi
Hi,

thanks for the response! Are there any plans to expose the intrinsic parameters calculated by the calibration in the future?

Also, this doesn't mean that the correct intrinsics aren't used when using the TT_* API, right? We are using NatNet to stream (unlabelled) marker data to our custom software for now, but I am thinking about switching to the more elaborate Motive SDK.

Re: Intrinsic parameters and camera orientation

Posted: Thu Nov 21, 2013 4:16 pm
by beckdo
Hey JeDi,

Exposing the complete extrinsic and intrinsic result calculated by the camera calibration has been a topic of ongoing discussion. It's not available currently but that's not to say it won't happen in the future. I've already referenced this thread in that regards to that topic internally.

The Motive API, including all of the TT_ methods do utilize the complete and correct calibration results from camera calibration they just don't expose the raw values externally.

Thanks,
Doug

Re: Intrinsic parameters and camera orientation

Posted: Fri Aug 02, 2019 4:39 pm
by Sandino
Hi guys,

I know that this is a very old post, but I want to know the state of retrieving the intrinsic parameters from the system guts. Are they available now? or maybe in a next release?

Re: Intrinsic parameters and camera orientation

Posted: Mon Aug 05, 2019 10:23 am
by steven.andrews
Hi Sandino,

Limited calibration information is accessible using the Motive API, but camera intrinsics are still not expose.
You can find a little more information in this Forum thread.

viewtopic.php?f=69&t=17247

Best,
Steven
--
Steven Andrews
OptiTrack | Senior Customer Support Engineer
help.naturalpoint.com
support@optitrack.com