Calibrate Camera
This command consists in using chessboard images to calibrate the camera. Calibrating camera means to compute internal parameters of the camera.
Requirements
No selection required to launch the command.
Set the different chessboard parameters (number of rows, lines and size of a square)
Import the images
Click Preview to see textual results:
Percentage of sensor surface where chessboard corners were found
Root mean square (RMS) error
Internal parameters
Activate/deactivate an image by using the icon next to the image name
Save the calibration results: the OK button is available when a calibration is done, but not available if an image is added or a point deleted: a new calibration has to be done by clicking on the Preview button to activate the OK button. When you push the OK button, the file is saved in the same folder as the images:
INCAM file: one for the project with all internal parameters. The .incam file is the most important because it can be used in all other projects with the same camera and focal length (to texture a mesh, for example); see Estimate Pose.
The INCAM file must be located in the same folder of the related image(s), so that the image(s) will be set to perspective type with the defined internal parameters from the INCAM. If there is no side file, the image type will be unknown unless it is an ortho-image with embedded world file.
Tips & Tricks
You can click on the button Display fullscreen chessboard (or press F11) to show the calibration chessboard in fullscreen according to the given parameters. This way, you do not have to print a calibration pattern on a piece of paper but you can directly take pictures of your screen.
Technical information
Internal parameters describe geometric elements of the camera and can then be used to texture automatically a photo taken from the same camera:
physical Size of sensor, in millimeters
size of a pixel on sensor in millimeters
focal length in millimeters
PPS: Principal Point of Symmetry, center of radial distortions. This point is relative to the center of sensor
radial distortions. Distortions represent geometric defects
The computation needs at least 3 pictures of a chessboard. On each image, the chessboard must be completely visible. All the images have to be taken by the same camera using the same focal length.