AI helps you reading Science

AI generates interpretation videos

AI extracts and analyses the key points of the paper to generate videos automatically


pub
Go Generating

AI Traceability

AI parses the academic lineage of this thesis


Master Reading Tree
Generate MRT

AI Insight

AI extracts a summary of this paper


Weibo:
We summarize the graph constructions used for regular functions

What energy functions can be minimized via graph cuts?

IEEE Transactions on Pattern Analysis and Machine Intelligence, no. 2 (2004): 147-159

Cited: 3239|Views413
EI WOS

Abstract

In the last few years, several new algorithms based on graph cuts have been developed to solve energy minimization problems in computer vision. Each of these techniques constructs a graph such that the minimum cut on the graph also minimizes the energy. Yet, because these graph constructions are complex and highly specific to a particular...More

Code:

Data:

0
Introduction
  • Introduction and overview

    Many of the problems that arise in early vision can be naturally expressed in terms of energy minimization.
  • The minimum cut in turn can be computed very efficiently by max flow algorithms
  • These methods have been successfully used for a wide variety of vision problems including image restoration [8, 9, 17, 19], stereo and motion [4, 8, 9, 18, 23, 27, 30, 31], voxel occupancy [34], multi-camera scene reconstruction [24] and medical imaging [5, 6, 22].
  • Two recent evaluations of stereo algorithms using real imagery with dense ground truth [32, 36] found that the best overall performance was due to algorithms based on graph cuts
Highlights
  • Introduction and overview

    Many of the problems that arise in early vision can be naturally expressed in terms of energy minimization
  • In this paper we focus on two classes of energy functions
  • We define the class F2 to be functions that can be written as a sum of functions of up to 2 binary variables at a time, E(x1, . . . , xn) = Ei + Ei,j
  • In section 2 we describe how graph cuts can be used to minimize energy functions, and discuss the importance of energy functions with binary variables in the context of two example vision problems, namely stereo and multi-camera scene reconstruction
  • We summarize the graph constructions used for regular functions
Conclusion
  • Summary of the results

    In this paper the authors focus on two classes of energy functions. Let {x1, . . . , xn}, xi ∈ {0, 1} be a set of binary-valued variables.
  • The authors define the class F3 to be functions that can be written as a sum of functions of up to 3 binary variables at a time, E(x1, .
  • The authors show how to construct a graph G for a regular function E(x1, x2) of two variables.
  • It will contain four vertices: V = {v1, v2, s, t}.
Related work
  • There is an interesting relationship between regular functions and submodular functions.7 Let S be a finite set and g : 2S → R be a real-valued function defined on the set of all subsets of S. g is called submodular if for any X, Y ⊂ S g(X) + g(Y ) ≥ g(X ∪ Y ) + g(X ∩ Y ).

    See [14], for example, for a discussion of submodular functions. An equivalent definition of submodular functions is that g is called submodular if for any X ⊂ S and i, j ∈ S − X g(X ∪ {j}) − g(X) ≥ g(X ∪ {i, j}) − g(X ∪ {i}).

    Obviously, functions of subsets X of S = {1, . . . , n} can be viewed as functions of n binary variables (x1, . . . , xn); the indicator variable xi is 1 if i is included in X and 0 otherwise. Then it is easy to see that the second definition of submodularity reduces to the definition of regularity. Thus, submodular functions and regular functions are essentially the same. We use different names to emphasize a technical distinction between them: submodular functions are functions of subsets of S while regular functions are functions of binary variables. From our experience, the second point of view is much more convenient for vision applications.
Funding
  • This research was supported by NSF grants IIS-9900115 and CCR-0113371, and by a grant from Microsoft Research
Reference
  • Ravindra K. Ahuja, Thomas L. Magnanti, and James B. Orlin. Network Flows: Theory, Algorithms, and Applications. Prentice Hall, 1993.
    Google ScholarFindings
  • Amir Amini, Terry Weymouth, and Ramesh Jain. Using dynamic programming for solving variational problems in vision. IEEE Transactions on Pattern Analysis and Machine Intelligence, 12(9):855–867, September 1990.
    Google ScholarLocate open access versionFindings
  • Stephen Barnard. Stochastic stereo matching over scale. International Journal of Computer Vision, 3(1):17–32, 1989.
    Google ScholarLocate open access versionFindings
  • S. Birchfield and C. Tomasi. Multiway cut for stereo and motion with slanted surfaces. In International Conference on Computer Vision, pages 489–495, 1999.
    Google ScholarLocate open access versionFindings
  • Yuri Boykov and Marie-Pierre Jolly. Interactive organ segmentation using graph cuts. In Medical Image Computing and Computer-Assisted Intervention, pages 276–286, 2000.
    Google ScholarLocate open access versionFindings
  • Yuri Boykov and Marie-Pierre Jolly. Interactive graph cuts for optimal boundary and region segmentation of objects in N-D images. In International Conference on Computer Vision, pages I: 105–112, 2001.
    Google ScholarLocate open access versionFindings
  • Yuri Boykov and Vladimir Kolmogorov. An experimental comparison of min-cut/max-flow algorithms for energy minimization in computer vision. In International Workshop on Energy Minimization Methods in Computer Vision and Pattern Recognition, volume 2134 of LNCS, pages 359–374. Springer-Verlag, September 2001.
    Google ScholarLocate open access versionFindings
  • Yuri Boykov, Olga Veksler, and Ramin Zabih. Markov Random Fields with efficient approximations. In IEEE Conference on Computer Vision and Pattern Recognition, pages 648–655, 1998.
    Google ScholarLocate open access versionFindings
  • Yuri Boykov, Olga Veksler, and Ramin Zabih. Fast approximate energy minimization via graph cuts. IEEE Transactions on Pattern Analysis and Machine Intelligence, 23(11):1222– 1239, November 2001.
    Google ScholarLocate open access versionFindings
  • W. H. Cunningham. Minimum cuts, modular functions, and matroid polyhedra. Networks, 15:205–215, 1985.
    Google ScholarLocate open access versionFindings
  • E. Dahlhaus, D. S. Johnson, C.H. Papadimitriou, P. D. Seymour, and M. Yannakakis. The complexity of multiway cuts. In ACM Symposium on Theory of Computing, pages 241–251, 1992. Revised version appears in the SIAM Journal of Computing, August 1994.
    Google ScholarLocate open access versionFindings
  • Jose Dias and Jose Leitao. The ZπM algorithm: A method for interferometric image reconstruction in sar/sas. IEEE Transactions on Image Processing, 11(4):408–422, April 2002.
    Google ScholarLocate open access versionFindings
  • L. Ford and D. Fulkerson. Flows in Networks. Princeton University Press, 1962.
    Google ScholarFindings
  • S. Fujishige. Submodular functions and Optimization, volume 47 of Annals of Discrete Mathematics. North Holland, 1990.
    Google ScholarFindings
  • S. Geman and D. Geman. Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images. IEEE Transactions on Pattern Analysis and Machine Intelligence, 6:721–741, 1984.
    Google ScholarLocate open access versionFindings
  • A. Goldberg and R. Tarjan. A new approach to the maximum flow problem. Journal of the Association for Computing Machinery, 35(4):921–940, October 1988.
    Google ScholarLocate open access versionFindings
  • D. Greig, B. Porteous, and A. Seheult. Exact maximum a posteriori estimation for binary images. Journal of the Royal Statistical Society, Series B, 51(2):271–279, 1989.
    Google ScholarLocate open access versionFindings
  • H. Ishikawa and D. Geiger. Occlusions, discontinuities, and epipolar lines in stereo. In European Conference on Computer Vision, pages 232–248, 1998.
    Google ScholarLocate open access versionFindings
  • H. Ishikawa and D. Geiger. Segmentation by grouping junctions. In IEEE Conference on Computer Vision and Pattern Recognition, pages 125–131, 1998.
    Google ScholarLocate open access versionFindings
  • Hiroshi Ishikawa. Exact optimization for markov random fields with convex priors. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2003. To appear.
    Google ScholarLocate open access versionFindings
  • Satoru Iwata, Lisa Fleischer, and Satoru Fujishige. A combinatorial, strongly polynomial-time algorithm for minimizing submodular functions. In ACM Symposium on Theoretical Computer Science, pages 97–106, 2000.
    Google ScholarLocate open access versionFindings
  • Junmo Kim, John Fisher, Andy Tsai, Cindy Wible, Alan Willsky, and William Wells. Incorporating spatial priors into an information theoretic approach for fMRI data analysis. In Medical Image Computing and Computer-Assisted Intervention, pages 62–71, 2000.
    Google ScholarLocate open access versionFindings
  • Vladimir Kolmogorov and Ramin Zabih. Visual correspondence with occlusions using graph cuts. In International Conference on Computer Vision, pages 508–515, 2001.
    Google ScholarLocate open access versionFindings
  • Vladimir Kolmogorov and Ramin Zabih. Multi-camera scene reconstruction via graph cuts. In European Conference on Computer Vision, volume 3, pages 82–96, 2002.
    Google ScholarLocate open access versionFindings
  • C.-H. Lee, D. Lee, and M. Kim. Optimal task assignment in linear array networks. IEEE Transactions on Computers, 41(7):877–880, 1992.
    Google ScholarLocate open access versionFindings
  • S. Li. Markov Random Field Modeling in Computer Vision. Springer-Verlag, 1995.
    Google ScholarFindings
  • Michael H. Lin. Surfaces with occlusions from layered stereo. PhD thesis, Stanford University, dec 2002.
    Google ScholarFindings
  • Ioannis Milis. Task assignment in distributed systems using network flow methods. In Combinatorics and Computer Science, volume 1120 of Lecture Notes in Computer Science, pages 396–405. Springer-Verlag, 1996.
    Google ScholarLocate open access versionFindings
  • Tomaso Poggio, Vincent Torre, and Christof Koch. Computational vision and regularization theory. Nature, 317:314–319, 1985.
    Google ScholarLocate open access versionFindings
  • S. Roy. Stereo without epipolar lines: A maximum flow formulation. International Journal of Computer Vision, 1(2):1–15, 1999.
    Google ScholarLocate open access versionFindings
  • S. Roy and I. Cox. A maximum-flow formulation of the n-camera stereo correspondence problem. In International Conference on Computer Vision, 1998.
    Google ScholarLocate open access versionFindings
  • Daniel Scharstein and Richard Szeliski. A taxonomy and evaluation of dense two-frame stereo correspondence algorithms. International Journal of Computer Vision, 47:7–42, April 2002.
    Google ScholarLocate open access versionFindings
  • Alexander Schrijver. A combinatorial algorithm minimizing submodular functions in strongly polynomial time. Journal of Combinatorial Theory, B80:346–355, 2000.
    Google ScholarLocate open access versionFindings
  • Dan Snow, Paul Viola, and Ramin Zabih. Exact voxel occupancy with graph cuts. In IEEE Conference on Computer Vision and Pattern Recognition, pages 345–352, 2000.
    Google ScholarLocate open access versionFindings
  • Harold S. Stone. Multiprocessor scheduling with the aid of network flow algorithms. IEEE Transactions on Software Engineering, pages 85–93, 1977.
    Google ScholarLocate open access versionFindings
  • Richard Szeliski and Ramin Zabih. An experimental comparison of stereo algorithms. In B. Triggs, A. Zisserman, and R. Szeliski, editors, Vision Algorithms: Theory and Practice, number 1883 in LNCS, pages 1–19, Corfu, Greece, September 1999. Springer-Verlag.
    Google ScholarFindings
0
Your rating :

No Ratings

Tags
Comments
avatar
数据免责声明
页面数据均来自互联网公开来源、合作出版商和通过AI技术自动分析结果,我们不对页面数据的有效性、准确性、正确性、可靠性、完整性和及时性做出任何承诺和保证。若有疑问,可以通过电子邮件方式联系我们:report@aminer.cn