I just found that I can get the FOV (field-of-view) parameters of the camera matrix from fovx, fovy, _, _, _ = cv2.calibrationMatrixValues(camera_matrix, resolution, *sensor_size).
My camera's sensor (OV3660) specifies in it's datasheet:
- resolution: 2048 px x 1536 px
- pixel size: 1.4 um x 1.4 um
- image area: 2912 um x 2167.2 um
People are often stating that for FOV calculation the pixel size is the measure to take. When I multiply my pixel size with the image resolution I ge a slightly smaller image area though. This is understandable, as there is probably some between the pixels material on the sensors physics.
Am I right to believe that the image area as specified for this sensor is the more accurate parameter to use?