Hi Erik,
If you need to measure planar objects with a single camera, then you need to know the camera extrinsics, which are the 3D rotation and translation between your plane of interest (table, conveyor belt, etc.) and the camera's coordinate system. If your camera does not move relative to that plane, then you only need to compute the extrinsics once. You can place a checkerboard on the plane once, compute the rotation and translation using the extrinsics function, save the results, and re-use them for measurements.
Alternatively, you can use a calibrated stereo pair of cameras. In that case, all the measurements will be in the coordinate system attached to one of the cameras, and you do not need a checkerboard in the scene.
