Great idea, nice job guys!  That is a much more elegant solution than our calibration target -- we used some aluminum squares sitting up on blocks (they're actually coasters I had sitting around, see attached).  We should probably decide on a standardized calibration target (glass with circles?) so we can all build one.

We also calibrated our rig today, and although our initial calibration still has a fair amount of error, we're starting to be able to get some reasonable looking XYZRGB point clouds: http://www.youtube.com/watch?v=YZ9bHf1CT2o .  With a more accurate calibration technique, it seems like the Kinect will be able to produce very nice data!

Cheers,
Alex


On Tue, Nov 16, 2010 at 9:35 PM, Radu Bogdan Rusu <rusu@willowgarage.com> wrote:
Hi Bill,

Great work! A few suggestions:

 * use a larger calibration pattern that covers more of the image

 * use a calibration pattern with more squares

 * if the estimates of the corners (after subpixel refinement) are still bad, an iterative fitting process (or LMedS or
RANSAC) could help


Alternatively, a calibration plate with circles would be better. We already have code ready that we need to commit for
finding out centers of circles/ellipses and using those for calibration plates a la
http://www.mvtec.com/halcon/applications/imgs/3d-vision-example-3d-calibration01.png, so this might improve things.


Cheers,
Radu.


On 11/16/2010 05:44 PM, Bill Morris wrote:
> Ivan had a great idea for using stereo calibration to calibrate between
> the depth image and the RGB by taking advantage of the optical
> properties of glass.
>
> So after a busy day of arts and crafts we would like to present the
> attached image of black squares rubber cemented to a sheet of glass.
>
> We should have more detailed results tomorrow.
>
> William Morris
> Ivan Dryanovski
> CCNY Robotics Lab
>
>
>
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
_______________________________________________
ros-users mailing list
ros-users@code.ros.org
https://code.ros.org/mailman/listinfo/ros-users