THE SPACE REQUIREMENTS OF INDEXING UNDER PERSPECTIVE PROJECTIONS

Object recognition systems can be made more efficient through the use of table lookup to match features. The cost of this indexing process d epends on the space required to represent groups of model features in such a lookup table. We determine the space required to perform indexi ng of arbitrary sets of 3-D model points for lookup from a single 2-D image formed under perspective projection. We show that in this case. one must use a 3-D surface to represent model groups, and we provide a n analytic description of such a surface. This is in contrast to the c ases of scaled-orthographic or affine projection, in which only a 2-D surface is required to represent a group of model features [3], [10]. This demonstrates a fundamental way in which the recognition of object s under perspective projection is more complex than is recognition und er other projection models.