Image analysis problems (e.g. segmentation and registration) are typically formulated in an optimization framework (energy-minimization). In this work, I present two important recurring questions: What to optimize and how to optimize and expose the resulting tradeoffs between the fidelity (of the domain and range) of the objective function and its optimizability.
The topic of this presentation was presented at: - Scandinavian Conference on Image Analysis, Ystad, Sweden (May 25, 2011). - Banff International Research Station workshop on Geometry for Anatomy, Canada (Sep 1, 2011). - University of British Columbia (UBC), Dept. Electrical and Computer Engineering (Oct 25, 2011). - The Workshop on Computational Aspects in Medical Imaging, MITACS International Focus Period on Mathematics of Medical Imaging, UBC, Canada (Oct 3, 2011)
The Optimizability-Fidelty Trade-Off in Image Analysis
Copyright is held by the author(s).
Member of collection