0

I have a camera setup with two or more cameras looking at he same object from opposite directions: enter image description here

I want to find a transformations between these cameras using asymmetric circles pattern and cv::stereoCalibrate. Is it going to work in this setup?

First problem that I've noticed is that OpenCV can't find inverted pattern, so second camera does not recognize pattern. To solve for that I flip image about x for second camera and then flip found points.

DikobrAz
  • 3,557
  • 4
  • 35
  • 53
  • I am surprised that using asymmetric circles pattern OpenCV cannot find the pattern for the second camera as it can detect chessboard pattern with any orientation. It should work with cv::stereoCalibrate. Other solution is to calibrate each camera independently, detect chessboard corners in the left and right images, compute the camera poses c1Mo, c2Mo for both cameras with cv::solvePnP and get the transformation c1Mc2=c1Mo*(c2Mo)^-1. – Catree May 17 '16 at 09:23
  • You right, I think I had different problem. Camera image was mirrored, so I had this weird issues when looking for chessboard pattern. – DikobrAz May 17 '16 at 17:19

0 Answers0