A Point Pattern Matching Algorithm Based on Minimize Spanning Tree and Fiedler Vector
Shanli Xuan, Dong Liang, Ming Bo Zhu, Yi-Zheng Fan, Nian Wang · 2009
Based on the minimize spanning tree and Fiedler vector, a new feature matching algorithm is proposed in this paper. Firstly, a weighted complete graph is constructed with the feature points of each image respectively, then search the minimal spanning tree in each complete graph. Secondly, perform spectral decomposition on the Gaussian-weighted Laplace matrix of the minimize spanning tree respectively, and divide the feature points into several different sets based on the vector (Fiedler vector) corresponding to the second smallest eigenvalue of Laplace matrix, then the corresponding relation between sets is obtained. And construct Gaussian-weighted Laplace matrices between the corresponding sets, then submit the matrices to spectral decomposition. Finally, complete the feature matching by constructing matching matrix with eigenvalues and eigenvectors. Experiment results indicate that the algorithm has a high accuracy.