Key points are not available for this paper at this time.
In this paper, we describe a technique for geometrically hash ing two-dimensional model objects. Used in conjunction with other methods for recognizing partially obscured and over lapping objects, this technique enables us to recognize over lapping, two-dimensional objects selected from large data bases of model objects without significant performance degradation when the database is enlarged. This technique is based on use of a synthetic attribute of an object, which we will call its footprint. Experimental results from databases of up to 100 objects are presented.
Kalvin et al. (Mon,) studied this question.