Camera Intrinsic calibration
- Click on ‘New camera intrinsic calibration’ to start the intrinsic calibration.
- This page also lists all saved calibrations and filters to view them.
- Start the process by clicking on ‘Get started'.
- Add the camera name and select the camera model(For wide-angle cameras, the Fish-eye model should be used)for which the intrinsic calibration has to be done.
- The target can be a checkerboard or a charuco board.
- The checkerboard target configuration defines the number of horizontal and vertical corners.
- For the charucoboard target, the user needs to configure the dictionary, square size and marker size along with the horizontal and vertical corners.
Camera lens model: For wide-angle cameras, the Fish-eye model should be used. For all other cameras, the Normal model can be used.
Horizontal corners: Number of horizontal inner corners in the checkerboard/ Number of chessboard squares in horizontal direction in charucoboard.
Vertical corners: Number of vertical inner corners in the checkerboard/ Number of chessboard squares in vertical direction in charucoboard.
Charucoboard Dictionary: There are multiple different types of the aruco markers from which the charuco board can be made. We support four commonly used dictionary of opencv which are DICT_4X4_250, DICT_5X5_250, DICT_6X6_250, DICT_7X7_250.
Square size: its the size of the each square in the board
Marker size: Its the size of the each marker present inside the charucoboard.
- Upload the images along with the checkerboard in its frame of view. Images are now limited to 50 per calibration, which means users can upload at most 50 images.
- On clicking on ‘Run calibration’ the intrinsic parameters are calibrated.
- One can verify the calibration results by using the reprojection error. It is the mean error of all the reprojected points.
- Also, the results can be visually verified by checking the undistorted image.
- The side by side view can be used to check both the distorted and undistorted image at the same time.
- Checkerboard coverage shows the amount of area that is covered by the checkerboard corners from all the images uploaded.
- Higher coverage results in higher calibration accuracy.
- Also users can see the individual reprojection error of all the checkerboard corner points. A color ramp is used to depict the reprojection error. A light red color shows the lower reprojection error, and the higher red color shows the higher reprojection error.
Camera Intrinsic parameters can be saved to profile can be saved using this button
Reprojection error is in pixels. It is the mean of the euclidean distance between the auto-detected checkerboard corners to the reprojected checkerboard corners. If the error is closer to zero it is better.