Efficient convex optimization for minimal partition problems with volume constraints
Möllenhoff T, Nieuwenhuis C, Töppe E, Cremers D (2013)
Publication Type: Conference contribution
Publication year: 2013
Journal
Book Volume: 8081 LNCS
Pages Range: 94-107
Conference Proceedings Title: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Event location: SWE
ISBN: 9783642403941
DOI: 10.1007/978-3-642-40395-8_8
Abstract
Minimal partition problems describe the task of partitioning a domain into a set of meaningful regions. Two important examples are image segmentation and 3D reconstruction. They can both be formulated as energy minimization problems requiring minimum boundary length or surface area of the regions. This common prior often leads to the removal of thin or elongated structures. Volume constraints impose an additional prior which can help preserve such structures. There exist a multitude of algorithms to minimize such convex functionals under convex constraints. We systematically compare the recent Primal Dual (PD) algorithm [1] to the Alternating Direction Method of Multipliers (ADMM) [2] on volume-constrained minimal partition problems. Our experiments indicate that the ADMM approach provides comparable and often better performance. © 2013 Springer-Verlag.
Involved external institutions
Germany (DE)How to cite
APA:
Möllenhoff, T., Nieuwenhuis, C., Töppe, E., & Cremers, D. (2013). Efficient convex optimization for minimal partition problems with volume constraints. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (pp. 94-107). SWE.
MLA:
Möllenhoff, Thomas, et al. "Efficient convex optimization for minimal partition problems with volume constraints." Proceedings of the 9th International Conference on Energy Minimization Methods in Computer Vision and Pattern Recognition, EMMCVPR 2013, SWE 2013. 94-107.
BibTeX: Download