A Comparison of Least Squares and Spectral Methods for Attributed Graph Matching
In this paper, least squares and spectral methods for attributed graph matching are compared. For the least squares method, complete graphs and decomposed graph models are considered in conjunction with the least squares approximations to optimal permutation matrices. We have used a version of Umeyama’s spectral method for comparison purposes. Results clearly demonstrate how both these methods are affected by additive noise but that, in general, least squares methods are superior.
KeywordsSpectral Method Input Graph Graph Match Adjacency Matrice Edge Attribute
- 3.Gold, S., Rangarajan, A.: A graduated assignment algorithm for graph matching. IEEE TPAMI 18(4), 377–388 (1996)Google Scholar
- 11.Grewe, L., Kak, A.C.: Interactive learning of a multiple-attribute hash table classifier for fast object recognition. CVIU 61(3), 387–416 (1995)Google Scholar