capabilities available in Map Enterprise and Descartes are "2D" to "2D" transformations as opposed to orthorectification that considers terrain model.
Several mathematic models are provided, you will find more details on those below (copy/paste from the help ):
The Helmert Model The Helmert model, also called “rigid body,” performs a rotation and two translations (X and Y); however, no scaling is done. This model guarantees that the uncorrected data retains its proportions since the image or design elements are only moved and rotated.
The Similitude Model The Similitude model performs a rotation, two translations (X and Y), and a scaling that has the same factor for the x- and y-axis. You can use a Similitude model to cut an image, to rotate an image, or to register two images with the same deformation relative to each other.
The Affine-1 Model The Affine-1 performs a rotation, two translations (X and Y), and applies a different scale factor for the x-, y-axis as well as a factor that changes the orthogonality of these axes. You can use the Affine-1 model to register a document that has very little deformation such as a scanned map.
The Projective Model The Projective model projects one plane into another plane. Like the Similitude and the Polynomial-1 models, it can be used to register a document that has very little deformation. For example, you can use it to register an aerial photograph that has little relief.
Thin Plate Spline Model The Thin Plate Spline model is based on a four variable mathematical algebraic formula. This means that this model requires at least four pair of control points to resolve the transformation model.
The benefits of this model resides in the fact that in order to compensate for the distortion caused by the rubbersheeting, instead of moving the control points to accommodate the residual values as in the traditional models, the Thin Plate Spline does not move the control points but instead, applies a correction elsewhere around the control points.
The trade-off is where it will most often be required to enter more control points than a traditional model. Control points will be positioned where the two systems (correct and uncorrected) do not aligned. Control points must be entered until the tanglement is satisfactory.
The Polynomial-2 and 3 models The Polynomial-2 and 3 models are used to register documents that have moderate relief. However, they are not suitable for documents that contain accentuated relief. For example, you can use these models to register aerial photographs of a city with low hills, but should not use these models to register aerial photographs of a mountainous region."
Hope this helps