From dense subgraph to graph matching: A label propagation approach

摘要: Graph matching (GM) is a fundamental problem in computer science, and it has been successfully applied to provide solutions many problems vision. In this paper, we consider GM as clustering an association graph whose nodes represent candidate correspondences between two graphs be matched. And take the dense subgraph good prior for correct correspondences, thus propose label propagation approach expand resolve whole cluster. The achieved by affinity-preserving manifold ranking algorithm with dynamic vector which enforces constraints. constraints introduced through doubly-stochastic normalization procedure. Extensive experiments demonstrate that our outperforms state-of-the-art algorithms especially presence of outliers deformation.