Many algorithms on meshes require the minimization of composite objectives, i.e., energies that are compositions of simpler parts. Canonical examples include mesh parameterization and deformation. We propose a second order optimization approach that exploits this composite structure to efficiently converge to a local minimum. Our main observation is that a convex-concave decomposition of the energy constituents is simple and readily available in many cases of practical relevance in graphics. We utilize such convex-concave decompositions to define a tight convex majorizer of the energy, which we employ as a convex second order approximation of the objective function. In contrast to existing approaches that largely use only local convexification, our method is able to take advantage of a more global view on the energy landscape. Our experiments on triangular meshes demonstrate that our approach outperforms the state of the art on standard problems in geometry processing, and potentially provide a unified framework for developing efficient geometric optimization algorithms.
%0 Journal Article
%1 Shtengel:2017:GOV:3072959.3073618
%A Shtengel, Anna
%A Poranne, Roi
%A Sorkine-Hornung, Olga
%A Kovalsky, Shahar Z.
%A Lipman, Yaron
%C New York, NY, USA
%D 2017
%I ACM
%J ACM Trans. Graph.
%K 2017 acm graphics paper siggraph
%N 4
%P 38:1--38:11
%R 10.1145/3072959.3073618
%T Geometric Optimization via Composite Majorization
%U http://doi.acm.org/10.1145/3072959.3073618
%V 36
%X Many algorithms on meshes require the minimization of composite objectives, i.e., energies that are compositions of simpler parts. Canonical examples include mesh parameterization and deformation. We propose a second order optimization approach that exploits this composite structure to efficiently converge to a local minimum. Our main observation is that a convex-concave decomposition of the energy constituents is simple and readily available in many cases of practical relevance in graphics. We utilize such convex-concave decompositions to define a tight convex majorizer of the energy, which we employ as a convex second order approximation of the objective function. In contrast to existing approaches that largely use only local convexification, our method is able to take advantage of a more global view on the energy landscape. Our experiments on triangular meshes demonstrate that our approach outperforms the state of the art on standard problems in geometry processing, and potentially provide a unified framework for developing efficient geometric optimization algorithms.
@article{Shtengel:2017:GOV:3072959.3073618,
abstract = {Many algorithms on meshes require the minimization of composite objectives, i.e., energies that are compositions of simpler parts. Canonical examples include mesh parameterization and deformation. We propose a second order optimization approach that exploits this composite structure to efficiently converge to a local minimum. Our main observation is that a convex-concave decomposition of the energy constituents is simple and readily available in many cases of practical relevance in graphics. We utilize such convex-concave decompositions to define a tight convex majorizer of the energy, which we employ as a convex second order approximation of the objective function. In contrast to existing approaches that largely use only local convexification, our method is able to take advantage of a more global view on the energy landscape. Our experiments on triangular meshes demonstrate that our approach outperforms the state of the art on standard problems in geometry processing, and potentially provide a unified framework for developing efficient geometric optimization algorithms.},
acmid = {3073618},
added-at = {2018-06-22T15:32:25.000+0200},
address = {New York, NY, USA},
articleno = {38},
author = {Shtengel, Anna and Poranne, Roi and Sorkine-Hornung, Olga and Kovalsky, Shahar Z. and Lipman, Yaron},
biburl = {https://www.bibsonomy.org/bibtex/2385f9cc28fb0b537f93f543a53dcb8e4/achakraborty},
description = {Geometric optimization via composite majorization},
doi = {10.1145/3072959.3073618},
interhash = {f5cd46b52cbc61371541c0b2fb9b6b84},
intrahash = {385f9cc28fb0b537f93f543a53dcb8e4},
issn = {0730-0301},
issue_date = {July 2017},
journal = {ACM Trans. Graph.},
keywords = {2017 acm graphics paper siggraph},
month = jul,
number = 4,
numpages = {11},
pages = {38:1--38:11},
publisher = {ACM},
timestamp = {2018-06-22T15:32:25.000+0200},
title = {Geometric Optimization via Composite Majorization},
url = {http://doi.acm.org/10.1145/3072959.3073618},
volume = 36,
year = 2017
}