Higher-order segmentation via multicuts

• Verfasst von: Kappes, Jörg Hendrik [VerfasserIn]
• Verfasst von: Speth, Markus [VerfasserIn]
• Verfasst von: Reinelt, Gerhard [VerfasserIn]
• Verfasst von: Schnörr, Christoph [VerfasserIn]
• Titel: Higher-order segmentation via multicuts
• Verf.angabe: Jörg Hendrik Kappes, Markus Speth, Gerhard Reinelt, Christoph Schnörr
• E-Jahr: 2016
• Jahr: [2016]
• Jahr des Originals: 2015
• Umfang: 16 S.
• Fussnoten: Available online 21 November 2015 ; Gesehen am 07.05.2020
• Titel Quelle: Enthalten in: Computer vision and image understanding
• Ort Quelle: San Diego, Calif. : Elsevier, 1995
• Jahr Quelle: 2016
• Band/Heft Quelle: 143(2016), Seite 104-119
• ISSN Quelle: 1090-235X
• DOI: doi:10.1016/j.cviu.2015.11.005
• Datenträger: Online-Ressource
• Sprache: eng
• Sach-SW: Correlation clustering
• Sach-SW: Discrete optimization
• Sach-SW: Global optimality
• Sach-SW: Higher-order graphical models
• Sach-SW: Image labeling
• Sach-SW: Markov random fields
• Sach-SW: Multicut
• Sach-SW: Partitioning
• Sach-SW: Polyhedral combinatorics
• Sach-SW: Segmentation
• K10plus-PPN: 1697664261

Abstract

Multicuts enable to conveniently represent discrete graphical models for unsupervised and supervised image segmentation, in the case of local energy functions that exhibit symmetries. The basic Potts model and natural extensions thereof to higher-order models provide a prominent class of such objectives, that cover a broad range of segmentation problems relevant to image analysis and computer vision. We exhibit a way to systematically take into account such higher-order terms for computational inference. Furthermore, we present results of a comprehensive and competitive numerical evaluation of a variety of dedicated cutting-plane algorithms. Our approach enables the globally optimal evaluation of a significant subset of these models, without compromising runtime. Polynomially solvable relaxations are studied as well, along with advanced rounding schemes for post-processing.