David Suter is a research professor at Edith Cowan University, Perth Western Australia. Prior to that he was a Professor as They University of Adelaide, South Australia (2008-2017), including serving as the Head of the School of Computer Science (2009-2013).
Before that he was Professor at Monash University (Melbourne, Australia). He is currently on the editorial board of the Journal Pattern Recognition, and previously served on the editorial boards of the International Journal of Computer Vision, amongst others.
He was general co-chair of ACCV2002 (Melbourne, Australia) and ICIP2013 (Melbourne, Australia).
Max consensus, implemented by RANSAC, has been the main work horse of computer vision algorithms requiring (robust) model fitting for several decades.
Despite being simple to implement, and generally reasonably effective, RANSAC comes with no performance guarantees except a weak probabilistic argument that it will give some sort of acceptable result if run long enough.
This has lead some researchers to look for deterministic ways to solve max consensus that are tractable and with improved guarantees on performance - including “half-way” house methods that deterministically update RANSAC to see a better solution.
This talk will outline recent methods in this vein.