A new framework is presented that uses tools from duality theory of linear programming to derive graph-cut based combinatorial algorithms for approximating NP-hard classification problems. The derived algorithms include /spl alpha/-expansion graph cut techniques merely as a special case, have guaranteed optimality properties even in cases where /spl alpha/-expansion techniques fail to do so and can provide very tight per-instance suboptimality bounds in all occasions.
A new framework for approximate labeling via graph cuts
Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1 ; 2 ; 1018-1025 Vol. 2
01.01.2005
296090 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
A New Framework for Approximate Labeling via Graph Cuts
British Library Conference Proceedings | 2005
|A new primal-dual algorithm for multilabel graph-cuts problems with approximate moves
British Library Online Contents | 2017
|A new primal-dual algorithm for multilabel graph-cuts problems with approximate moves
British Library Online Contents | 2017
|British Library Online Contents | 2012
|Illumination-invariant tracking via graph cuts
IEEE | 2005
|