Some cameras you might get and would like to use for photogrammetry come with all the required info in manuals, metafiles or have already been calibrated by someone else. But what happens if you are missing something? Focal length, etc is relatively easy to get but the lens distortions (radial, tangential) are not. We had an aerial survey coming up and needed to get a camera calibrated ASAP. When you search online expect to get out of your depth rapidly in matrices and formulae. But then if you are lucky you might stumble on OpenCV.
OpenCV is an open source computer vision and machine learning library which has what you need to calculate the distortion coefficients. As usual the most complicated thing is getting the software to install, run or compile so I ended up pulling what I needed out and running it in Python separately. In theory it seems to work and finally I have a legitimate reason for having a chessboard up on my monitor. It also means I can start trying to find the other values I need for the camera.
Edit: That is not me, my office or my computer!
No comments:
Post a Comment