An Efficient Geometric Constraint Method for Robust Local Feature Matching


Local feature matching is not robust to extract correct correspondences under many conditions, such as images with general deformations and repetitive patterns. To solve this problem, this paper proposes a new geometric constraint method: Maximal Clique Matching (MCM). In MCM, the global geometric constraint problem can be expressed as the maximal clique problem in graph theory. MCM starts from building a geometric correspondence graph (GCG) based upon the pairwise geometric information in local features, and then an efficient heuristic approximation algorithm is developed to get the global geometric relationships by finding the maximal cliques in GCG. Given the characteristics of the global optimality of maximal cliques, MCM is robust to occlusion, clutter, deformations and repetitive patterns. We evaluated the method using two public datasets. Results show that our method outperforms other up-to-date techniques.

The 12th IAPR Conference on Machine Vision Applications
Li Song
Li Song
Professor, IEEE Senior Member

Professor, Doctoral Supervisor, the Deputy Director of the Institute of Image Communication and Network Engineering of Shanghai Jiao Tong University, the Double-Appointed Professor of the Institute of Artificial Intelligence and the Collaborative Innovation Center of Future Media Network, the Deputy Secretary-General of the China Video User Experience Alliance and head of the standards group.