Camera Calibration or Projection Mapping

Hello everyone,

I have this scenario, which seems pretty uncommon.

If anyone can tell me how to do camera mapping or camera projection (whatever it is called), as we can do in 3Ds max. What I want is, to create a 3D object, say a cube, which should match exactly to a 2D box in the background picture.

I am not sure if it has anything to do with Camera or does projection of 3D space has to be changed?

What I am doing so far:

I have a Orthographic Camera, which is only looking to the Background image Then I have a Perspective camera which will render everything else.

Now I want to map basic objects (i.e. cube or sphere for now) to the box shown in 2D image.

Any help in this regard will be highly appreciated.

Thanks.