Results of camera calibration vary

edit

Thank you very for your advices. I switched to OpenCV from JavaCV to try out with the original calibration demo. Now I'm playing with it and hope will get better results. Just small question it is not important to place the chessboard far away from camera? You said that it is important to cover as much as possible of the image with the chessboard I thought the difference in position(depth) are making the results better:-/

Well, I am still learning as well, so if you find a trustworthy source that tells you that far away chessboards are important, listen to them first. But if you pull up the far away chessboards intermeadiate results (the red/rainbow circles and lines drawn on the frames), you'll see a ton of error on low-res, if you get detects at all. You could do a bunch of these and pick out the good ones. However, my intution, based on my moderate-level understanding of what's going on under the hood is that for the "main" camera cal matrix parameters, fx, fy, cx, and cy, should be OK with only big chessboard views (make sure to remember to tilt/turn). I think the smaller views might help pick out lens and translation distorion.

Sounds reasonable. With chessboard closer to the camera the result seems to be better. I'm still having big avg.backprojection error, so I have some more questions please: 1) the cx,cy values are not very often referring to the exact middle of the sensor, is there any way how I can find them and use them as fixed afterwards? 2)What exactly is the "CV_CALIB_SAME_FOCAL_LENGTH" is it just to simplify the calculation of the calibration? 3)Is it at all possible to get backprojection errors really close to zero with such a bad camera? 4) can I somehow check manually the correctness of the intrinsic matrix? Let say I will use the points on the chessboard using real size in mm and than for known distance of some point from the camera calculate it's projection and compare them with real picture?

1. cx & cy are the point that the camera's principal axis intersects the image frame. In a perfect camera, they would be in the middle. In a cheap webcam, if they were in the middle it would be pure luck,
2. I think "CV_CALIB_SAME_FOCAL_LENGTH" just means that all the test chessboards were taken with the camera at the same focal length. I've never messed around with this off, but I'd imagine it lets you calibrate with images of chessboards taken with different focal lengths - probably by adding some more information as well.
3. I don't have the error info on me, but the program is telling me I am getting less than 2 pixels of error with a pretty horrible camera.

1. Possibly. Going directly into the camera matrix is hard, if not impossible. The focal lengths fx, and fy are actually a combination (multiplication?) of the physical focal length of the optical lens and the length and width of the individual CMOS sensor elements. The units on a camera matrix are some weird pixel/mm thing. You might be able to indirectly calculate based on multiple views of a known object from known distances or something - but then you're really doing the same sort of thing as the OpenCV anyways.