With more time, it would have been more robust to have used a system such as the Motion Analysis method. However, the Motion Analysis method returns motion information, so the camera has to be matched on the first frame. The Planar Mapping system, derived in this paper, positions the camera relative to the object it is tracking. So assuming the scale given is accurate then the position calculated for the camera does not require matching.
The tracking algorithm is the weaker link in the process, as it stands it doesn't deal with sub-pixel accuracy. This results in jittery representation of the motion displayed in the image sequence.