Multi camera calibration with flat pattern, error minimization.

asked 2019-09-09 03:04:15 -0500

Bleach gravatar image

I have dataset from camera array with chessboard markers. For simplicity, assume that have 3 cameras. To calibrate I use calibrateCamera alternately for each camera to find intrinsic on all images where pattern was found. After I use stereoCalibrate with fixed intrinsic on all images with same timestamp where pattern found. stereoCalibrate was performed to cameras 1-2 and 1-3. Now I have related position and orientation camera1 to cam2 and cam3. Please advice me optimization extrinsic method to all cameras together. As I understood bundle adjustment is not suitable for flat pattern, "depth" needed? Thanks.

edit retag flag offensive close merge delete



For the 3 camera case you may want to look into using the trifocal tensor instead of two stereo calibrations. I'm not aware of openCV support for trifocal tensor, though. (See Multiple View Geometry, Hartley / Zisserman for background information) This might get you the optimized results you want, for 3 cameras.

What error are you trying to minimize, specifically?

swebb_denver gravatar imageswebb_denver ( 2019-09-11 15:39:29 -0500 )edit

Thanks. I wrote 3 cameras for simplicity. Now there are 4, and maybe later there will be more. I try to minimize extrinsic reprojection errors on all cameras together.

Bleach gravatar imageBleach ( 2019-09-13 05:22:21 -0500 )edit