A Path Following Algorithm for Graph Matching
We propose a convex-concave programming approach for the labelled weighted graph matching problem. The convex-concave programming formulation is obtained by rewriting the graph matching problem as a least-square problem on the set of permutation matrices and relaxing it to two different optimization problems: a quadratic convex and a quadratic concave optimization problem on the set of doubly stochastic matrices. The concave relaxation has the same global minimum as the initial graph matching problem, but the search for its global minimum is aslo a complex combinatorial problem. We therefore construct an approximation of the concave problem solution by following a solution path of the convex-concave problem obtained by linear interpolation of the convex and concave formulations, starting from the convex relaxation. The algorithm is compared with some of the best performing graph matching methods on three datasets: simulated graphs, QAPLib and handwritten chinese characters.
Unable to display preview. Download preview PDF.
- 1.Filatov, A., Gitis, A., Kil, I.: Graph-based handwritten digit string recognition. In: Third International Conference on Document Analysis and Recognition (ICDAR 1995), pp. 845–848 (1995)Google Scholar
- 3.Schellewald, C., Schnor, C.: Probabilistic subgraph matching based on convex relaxation. Lecture notes in computer science, pp. 171–186. Springer, Heidelberg (2005)Google Scholar
- 5.Umeyama, S.: An eigendecomposition approach to weighted graph matching problems. Transaction on pattern analysis and machine intelligence 10 (1988)Google Scholar
- 7.Caelli, T., Kosinov, S.: An eigenspace projection clustering method for inexact graph matching. TPAMI 24 (2004)Google Scholar
- 8.Almohamad, H., Duffuaa, S.O.: A linear programming approach for the weighted graph matching problem. TPAMI 15 (1993)Google Scholar
- 9.Gold, S., Rangarajan, A.: A graduated assignment algorithm for graph matching. Transaction on pattern analysis and machine intelligence 18 (1996)Google Scholar
- 10.Cremers, D., Kohlberger, T., Schnor, C.: Evaluation of convex optimization techniques for the graph-matching problem in computer vision. Patter Recogn. 2191 (2001)Google Scholar
- 11.Belongie, S., Malik, J., Puzicha, J.: Shape matching and object recognition using shape contexts. Transaction on pattern analysis and machine intelligence 24 (2002)Google Scholar
- 13.Zaslavskiy, M., Bach, F., Vert, J.P.: Path following algorithm for graph matching problem. arXiv:0801.3654v1 (2008)Google Scholar
- 15.Chung, F.R.K.: Spectral Graph Theory. Americal Mathematical Society (1997)Google Scholar
- 17.Newman, M.E.J., Strogatz, S.H., Watts, D.J.: Random graphs with arbitrary degree distributions and their applications. PHYSICAL REVIEW 64 (2001)Google Scholar
- 18.Cela, E.: Qaudratuc assignment problem library (2007), http://www.opt.math.tu-graz.ac.at/qaplib/