Graph matching is an important problem in computer vision and pattern recognition. This paper proposes a graph matching algorithm based on a directed graph structure model. Firstly, the point set is divided into two parts by using the line between two points, and the direction information between the two points is specified according to the number of points on both sides of the line. Secondly, the ratio between the number of points on both sides of the line and the total number of points is used to define the adjacency matrix. Finally, the matching matrix is solved by the convex-concave relaxation procedure method to complete the matching. The invariance of the directed graph model under affine transformation is proved. The experimental results show that the directed graph model constructed can effectively and accurately describe the structure information between feature points and the matching method as well as has a high performance.

图匹配是计算机视觉和模式识别中的重要问题。提出了一种基于有向图结构模型的图匹配算法。首先，通过使用两点之间的直线将点集分为两部分，并根据直线两侧的点数指定两点之间的方向信息。其次，将线两侧的点数与总点数之比用于定义邻接矩阵。最后，利用凸凹松弛过程法求解匹配矩阵，完成匹配。证明了仿射变换下有向图模型的不变性。