Hi Eli
It is my understanding that you would like to use pixel to angle measurements for camera calibration instead of chess board.
Calibrating a camera with pixel-to-angle measurements is unconventional and lacks the geometric features offered by a chessboard pattern. Yet, if the angles are accurately known and cover the entire field of view, it may be possible to estimate certain intrinsic parameters.
Below are the conceptual steps that can be considered for the calibration:
- Import Data: Load the pixel to angle data from the Excel file into MATLAB.
- Define Camera Model: Specify the pinhole camera model with unknown focal length, principal point, and skew. Include a distortion model with radial (k1, k2, k3) and tangential (p1, p2) distortion coefficients.
- Establish Relationships: Create a mathematical relationship between the pixel positions and the corresponding angles.
- Optimization: Provide initial guesses for the camera parameters. Define an error function that represents the difference between measured angles and model predictions. Use an optimization algorithm to adjust the parameters and minimize the error function.
- Validation: Test the accuracy of the estimated parameters against additional measured data.
- Save Parameters: Save the calibrated camera parameters for future use.
You can refer to the following MathWorks documentation links to know more about "cameraParameters" and "lsqnonlin":
Hope this helps!
