Probabilistic correlation clustering and image partitioning using perturbed multicuts

• Verfasst von: Kappes, Jörg Hendrik [VerfasserIn]
• Verfasst von: Swoboda, Paul [VerfasserIn]
• Verfasst von: Savchynskyy, Bogdan [VerfasserIn]
• Verfasst von: Schnörr, Christoph [VerfasserIn]
• Titel: Probabilistic correlation clustering and image partitioning using perturbed multicuts
• Verf.angabe: Jörg Hendrik Kappes, Paul Swoboda, Bogdan Savchynskyy, Tamir Hazan, Christoph Schnörr
• Umfang: 12 S.
• Fussnoten: Gesehen am 07.03.2019
• Titel Quelle: Enthalten in: Scale Space and Variational Methods in Computer Vision
• Jahr Quelle: 2015
• Band/Heft Quelle: (2015), S. 231-242
• ISBN Quelle: 978-3-319-18461-6
• DOI: doi:10.1007/978-3-319-18461-6_19
• Datenträger: Online-Ressource
• Sprache: eng
• K10plus-PPN: 1588417395

Abstract

We exploit recent progress on globally optimal MAP inference by integer programming and perturbation-based approximations of the log-partition function. This enables to locally represent uncertainty of image partitions by approximate marginal distributions in a mathematically substantiated way, and to rectify local data term cues so as to close contours and to obtain valid partitions. Our approach works for any graphically represented problem instance of correlation clustering, which is demonstrated by an additional social network example.