DocumentCode :
3549194
Title :
Energy minimization via graph cuts: settling what is possible
Author :
Freedman, Daniel ; Drineas, Petros
Author_Institution :
Dept. of Comput. Sci., Rensselaer Polytech. Inst., Troy, NY, USA
Volume :
2
fYear :
2005
fDate :
20-25 June 2005
Firstpage :
939
Abstract :
The recent explosion of interest in graph cut methods in computer vision naturally spawns the question: what energy functions can be minimized via graph cuts? This question was first attacked by two papers of Kolmogorov and Zabih, in which they dealt with functions with pair-wise and triplewise pixel interactions. In this work, we extend their results in two directions. First, we examine the case of k-wise pixel interactions; the results are derived from a purely algebraic approach. Second, we discuss the applicability of provably approximate algorithms. Both of these developments should help researchers best understand what can and cannot be achieved when designing graph cut based algorithms.
Keywords :
computer vision; functions; graph theory; minimisation; approximate algorithm; computer vision; energy function minimization; graph cut method; k-wise pixel interaction; Algorithm design and analysis; Application software; Computer science; Computer vision; Explosions; Minimization methods; NP-complete problem; Polynomials; Stereo vision; Sufficient conditions; energy minimization; graph cuts;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Computer Vision and Pattern Recognition, 2005. CVPR 2005. IEEE Computer Society Conference on
ISSN :
1063-6919
Print_ISBN :
0-7695-2372-2
Type :
conf
DOI :
10.1109/CVPR.2005.143
Filename :
1467543
Link To Document :
بازگشت