The Tenth IASTED International Conference on
Computer Graphics and Imaging
~CGIM 2008~
February 13 – 15, 2008
Innsbruck, Austria
SPECIAL SESSION
Invariance And Robustness
|
Alex Balinsky Cardiff University, UK |
Jan Flusser Institute of Information Theory and Automation, Czech Republic |
|
Boris Kovalerchuk Central Washington University, USA |
Peter Revesz University of Nebraska, USA |
Abstract
Invariants play an important role in Computer Vision and their study has a long history. Different periods had different focuses, but the global goal was and still is the development of algorithms that will produce the same results for significantly different images.
The concept of significantly different images is in the permanent change to accommodate more and more differences in images emerging in growing medical, geospatial, industrial and other applications. The assumptions of the same scene, the same optical sensor, the same lighting and noise conditions coupled with assumptions that sensors are at distant locations from the scene put forward many studies of invariance relative to affine transformations. Several important affine invariants have been found and used in design of the pattern recognition, image registration and conflation algorithms under these assumptions.
The need to weaken assumptions about images poses new challenges and leads to many open research questions such as:
- How robust are knows affine and other invariants when assumptions are weakened?
- How to design new invariants?
- How far we can weaken assumptions to keep known invariants useful?
- How to design new invariants that are robust and can be used in design of pattern recognition algorithms?
- How to define robustness of invariants and algorithms based on invariants?
- Are there other groups of transformations beyond the affine group for which robust invariants exist?
Preliminary studies had shown that many known invariants used in pattern recognition algorithms are not robust to small changes in images and robust parameters of these algorithms are not invariant. The list of questions is not complete and papers that address other relevant questions are encouraged.
The goal of the special session is to set up a research agenda for studies on invariance and robustness. Both technical and position papers are invited. At the end of the session, a round table discussion of future directions in this research area will be held. For preparing papers general conference instructions should be used.
Papers should be submitted to one of the session organizers: BalinskyA@cardiff.ac.uk, flusser@utia.cas.cz, borisk@cwu.edu, revesz@cse.unl.edu.
Special Session Organizers
Alex Balinsky, Cardiff University, UK
Jan Flusser, Institute of Information Theory and Automation, Czech Republic
Boris Kovalerchuk, Central Washington University, USA
Peter Revesz, University of Nebraska, USA
Important Deadlines * Deadlines Extended *
| Submissions due | November 1, 2007 |
| Notification of acceptance | November 21, 2007 |
| Final manuscripts due | December 3, 2007 |
| Registration deadline | December 10, 2007 |
