Partial enumeration and curvature regularization Conference Paper

Author(s): Olsson, Carl; Ulen, Johannes; Boykov, Yuri; Kolmogorov, Vladimir
Title: Partial enumeration and curvature regularization
Affiliation IST Austria
Abstract: Energies with high-order non-submodular interactions have been shown to be very useful in vision due to their high modeling power. Optimization of such energies, however, is generally NP-hard. A naive approach that works for small problem instances is exhaustive search, that is, enumeration of all possible labelings of the underlying graph. We propose a general minimization approach for large graphs based on enumeration of labelings of certain small patches. This partial enumeration technique reduces complex high-order energy formulations to pairwise Constraint Satisfaction Problems with unary costs (uCSP), which can be efficiently solved using standard methods like TRW-S. Our approach outperforms a number of existing state-of-the-art algorithms on well known difficult problems (e.g. curvature regularization, stereo, deconvolution); it gives near global minimum and better speed. Our main application of interest is curvature regularization. In the context of segmentation, our partial enumeration technique allows to evaluate curvature directly on small patches using a novel integral geometry approach.
Conference Title: ICCV: International Conference on Computer Vision
Conference Dates: December 1-8, 2013
Conference Location: Sydney, Australia
Publisher: IEEE  
Date Published: 2013-01-01
Start Page: 2936
End Page: 2943
Sponsor: This work has been funded by the Swedish Research Council (grant 2012-4213), the Crafoord Foundation, the Canadian Foundation for Innovation (CFI 10318) and the Canadian NSERC Discovery Program (grant 298299-2012RGPIN).
DOI: 10.1109/ICCV.2013.365
Open access: yes (repository)
IST Austria Authors
Related IST Austria Work