Camera Calibration and Correction of Distortion

The CameraCalibrator object is used for camera calibration and correction of image distortion. It can be created by calling FacerecService.createCameraCalibrator.

To correct the distortion, you must either perform a calibration or just load the calibrated parameters using the CameraCalibrator.loadCameraParameters member function. After that, you can correct the image distortion using the CameraCalibrator.undistort member function.

For calibration you need to print a calibration template and fix it on a flat surface, preferably black.

There are three pattern types:

• asymmetric circles grid pattern
• chessboard pattern
• circles grid pattern

Here are examples of the patterns:

Full size images are stored in the share/calibration directory.

We recommend you to use the asymmetric circles grid pattern.

Before calibration you have to initialize the CameraCalibrator object by calling the CameraCalibrator.initCalibration member function. Then, you have to capture the pattern in different poses with the camera you want to calibrate and provide the captured frames in CameraCalibrator.addImage.

You can get the approximate progress of pattern pose space coverage by the CameraCalibrator.getPatternSpaceCoverProgress member function. You can get a tip what pattern pose is required using the CameraCalibrator.getTip member function.

Then, call the CameraCalibrator.calibrate method.

After a successful calibration, you can see the result by calling the CameraCalibrator.undistort method. Then you can save the calibrated parameters using the CameraCalibrator.saveCameraParameters method, if the result suits you, or add more images and perform a calibration again.

To reset the calibration, call the CameraCalibrator.initCalibration method again.