Fisheye Calibration using Opencv _ MUST_HAVE_

  1. https://github.com/LifeBeyondExpectations/fisheye-calibration-opencv-python-ROS
  2. You must upgrade your opencv version to 3.3.0 or your code get error for cv2.fisheye.calibrate( … )
  3. There are some problems for cv2.fisheye.CALIB_CHECK_COND
    At first, I got 200 results of objectpoints and imagepoints for fisheye calibration but did not have good results. Fig 01 is one result.
Fig 01. Results of the fisheye calibration using more than 200 images. It is hardly to get good result if you have too many images to have optimal camera matrix using cv2.fisheye.calibration
Fig 02. One example of checkerboard images. This kind of images are used for fisheye calibration.

3. I reduce the number of checkerboard images to 100 and still fail and got relatively good results with 40 checkerboard images. I do not know the exact reasons but I hope you got trouble in fisheye calibration using python and opencv, please do not have too much checkboard images for calibration.

4. Final Result are below.

Fig 03. Results of the fisheye calibration using opencv. Left column images are original frames. and the Right images are the results. Still, it needs to be fixed in detail.