VASC Seminar Announcement
            =========================

Date: Monday, 4/19/99
Time: 3:30-4:30
Place: Smith Hall 2nd Floor Common Area

Speaker:  Michael Greenspan
          Carleton University and National Research Council of Canada


Title: Geometric Probing in Dense Range Data

Abstract:

A method is presented which uses a variant of Geometric Probing to solve a 
general version of the 3D Object Recognition problem in dense range data. 

Geometric probing is a minimalist technique for object recognition which has 
evolved as a problem of  Computational Geometry. The presented method 
iteratively queries a range image for the existence of surface data at specific points. Depending upon the result of a query, the method reduces the set of 
possible interpretations of the scene, and adapts the strategy for the next 
probe. The probe strategy is very naturally represented as a Decision Tree 
Classifier.

The problem addressed is a very general version of the model-based 3D object 
recognition problem. There are no restrictions on the position or shape of the 
object under scrutiny. The imaged scenes can be cluttered and occluded. The 
method is robust to image noise, which is modeled via the "fallible oracle 
assumption". Results are presented on real images using a variety of 
combinations of decision tree design heuristics and tree traversal methods. The 
method is shown to be analogous to the efficient matching of a template set.


Biography

Michael Greenspan has been working with the Visual Information Technology Group 
of the National Research Council of Canada since 1992. His interests include 
object and pattern recognition, motion planning, robotics, and parallel 
computing, and holds several patents related to these fields. Michael received 
a B.Sc. in Physics and Applied Mathematics from the University of Toronto in 
1986, and the B.A.Sc. and M.A.Sc. degrees in Electrical Engineering from the 
University of Ottawa in 1989 and 1991. He is currently a Ph.D. candidate in the 
Systems and Computer Engineering department at Carleton University.