Multi-layer Random Walks Synchronization for Multi-attributed Multiple Graph Matching
Many applications in computer vision can be formulated as a multiple graph matching problem that finds global correspondences across a bunch of data. To solve this problem, matching consistency should be carefully considered with matching accuracy to prevent conflicts between graph pairs. In this paper, we aim to solve a multiple graph matching problem in complicated environments by using multiple attributes that are represented in a set of multi-layer structures. The main contribution of this paper is twofold. First, we formulate the global correspondence problem of multi-attributed graphs using multiple layered structures. The formulation is derived by aggregating the multi-layer structures that describe individual pairwise matching problems respectively. Second, we solve the global correspondence problem by using a novel multi-attributed multiple graph matching method that is based on the multi-layer random walks framework. The proposed framework contains additional synchronization steps to lead random walkers to consistent matching candidates. In our extensive experiments, the proposed method exhibited robust and accurate performance over the state-of-the-art algorithms.
READ FULL TEXT