home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!concert!borg!hatteras!boggs
- From: boggs@hatteras.cs.unc.edu (Donna Boggs)
- Newsgroups: triangle.talks
- Subject: UNC: CS Department Colloquium
- Message-ID: <16363@borg.cs.unc.edu>
- Date: 14 Oct 92 20:08:41 GMT
- Sender: news@cs.unc.edu
- Lines: 52
-
-
- THE UNIVERSITY OF NORTH CAROLINA
- AT
- CHAPEL HILL
-
- COMPUTER SCIENCE DEPARTMENT
-
- TWO TALKS ON IMAGE ANALYSIS
-
-
- DEPARTMENT COLLOQUIUM
-
- ALAN COLCHESTER, M.D., Ph.D.
- Guys Hospital, University of London
-
- A Unified Approach to the Segmentation of Grey-Level and Dot-Pattern Images
-
- Monday, October 19, 1992
- Rm. 011 Sitterson Hall at 3:30
-
-
-
- Informal Talk: GABOR SZEKELY, Ph.D.
- Communications Techology Laboratory
- ETH, Zurich, Switzerland
-
- Title: The Structure of the 3D Voronoi Diagram
- Used in Finding a Medial Surface of a 3D Object
-
- Where: Monday, October 19, 1992 at Image Lunch at 12:00
- Conference Area 284, Sitterson Hall
-
- Abstract: The Medial Axis Transformation was defined by Blum as a
- continuous concept. Its implementation on a discrete raster suffers under
- the requirement of agreeing with both a Euclide angeometry and a topology
- based on a regular metric. One elegant way to resolve this contradiction
- is using the Voronoi diagram of the boundary points as a superset of the
- skeleton. The microscopic details on the boundary produce a vast amount
- of 'noise branches' which should be pruned by appropriate algorithms. The
- known pruning algorithms can be showed to be based on a recursive scheme
- which relies heavily on the hierarchical branched structure of the 2D
- Voronoi skeleton. When generalizing the skeleton generation to 3D, one can
- try to solve the pruning problem by the same hierarchical approach. However,
- due to the extra degree of freedom, the structure of the Voronoi diagram has
- a much higher connectedness, resulting in heavily aggregated structures
- instead of treelike branching. The talk will overview our present understanding
- about the structure of 2D and 3D Voronoi diagrams and discuss the present
- difficulties.
-
- Host: Dr. Stephen M. Pizer, 2-1768. Refreshments will be served at 4:30
-
-
-