since the transformation is affine and not perspective it is more efficient to use the warpAffine function