Efficiently solving dynamic Markov random fields using graph cuts

In this paper, we present a fast new fully dynamic algorithm for the st-mincut/max-flow problem. We show how this algorithm can be used to efficiently compute MAP estimates for dynamically changing MRF models of labeling problems in computer vision, such as image segmentation. Specifically, given the solution of the max-flow problem on a graph, we show how to efficiently compute the maximum flow in a modified version of the graph. Our experiments showed that the time taken by our algorithm is roughly proportional to the number of edges whose weights were different in the two graphs. We test the performance of our algorithm on one particular problem: the object-background segmentation problem for video and compare it with the best known st-mincut algorithm. The results show that the dynamic graph cut algorithm is much faster than its static counterpart and enables real time image segmentation. It should be noted that our method is generic and can be used to yield similar improvements in many other cases that involve dynamic change in the graph

[1]  E. A. Dinic.  Algorithm for solution of a problem of maximal flow in a network with power estimation , 1970 .

[2]  D. Greig,et al.  Exact Maximum A Posteriori Estimation for Binary Images , 1989 .

[3]  R. Tamassia,et al.  Dynamic expression trees and their applications , 1991, SODA '91.

[4]  Olga Veksler,et al.  Markov random fields with efficient approximations , 1998, Proceedings. 1998 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Cat. No.98CB36231).

[5]  Olga Veksler,et al.  Fast approximate energy minimization via graph cuts , 2001, Proceedings of the Seventh IEEE International Conference on Computer Vision.

[6]  Marie-Pierre Jolly,et al.  Interactive Graph Cuts for Optimal Boundary and Region Segmentation of Objects in N-D Images , 2001, ICCV.

[7]  Y.Y. Boykov,et al.  Interactive graph cuts for optimal boundary & region segmentation of objects in N-D images , 2001, Proceedings Eighth IEEE International Conference on Computer Vision. ICCV 2001.

[8]  Mikkel Thorup,et al.  Fully-dynamic min-cut , 2001, STOC '01.

[9]  Hiroshi Ishikawa,et al.  Exact Optimization for Markov Random Fields with Convex Priors , 2003, IEEE Trans. Pattern Anal. Mach. Intell..

[10]  Vladimir Kolmogorov,et al.  What energy functions can be minimized via graph cuts? , 2002, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[11]  Vladimir Kolmogorov,et al.  An experimental comparison of min-cut/max- flow algorithms for energy minimization in vision , 2001, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[12]  Patrick Pérez,et al.  Interactive Image Segmentation Using an Adaptive GMMRF Model , 2004, ECCV.

[13]  Andrew Zisserman,et al.  OBJ CUT , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).

[14]  X. Liy.  Dynamic Algorithms in Computational Geometry , 2007 .