The transformation is derived from the homography between the reference surface coordinate system and the target image coordinate system, allowing for the projection of the 3D model into the image’s pixel space.