A system for automatically detecting targets is being developed using
their invariant geometrical properties. In order to overcome the
difficulties involved in building such a system, namely varying viewpoints,
changes, articulation and clutter in scenes and inevitable errors in low-level
image processing, the approach employed involves:
Restriction to a set of models (Model-based)
Feasible for man-made targets.
Overcomes the limitations of low-level image processing.