In recent years, the graph-based approach has become increasingly popular for modeling real-world data because graphs represent a wide variety of data, such as social networks and trajectory mining. In this paper, we propose a Weighted Adjacency Matrix (WAM) for constructing a graph based on edges from the features of each trajectory and a vertex from the frame number. In trajectory mining, finding objects that have similar patterns of motion is a common data analysis; thus, selecting effective algorithms is necessary for grouping information and showing trajectories as graphs. A new algorithm is proposed to calculate Graph Trajectory Similarity (GTS) between two trajectories of graphs, the Nodes Similarity Function (NSF). The NSF is introduced in our model for comparing similarity between two trajectories of graphs. A minimum spanning tree (MST) is a new schema used to prune edges and nodes of a graph.
Construction of a Graph of Similar Trajectories Based on the Minimum Spanning Tree (MST) and Node Similarity Function (NSF)
Computer Vision (CV)
Video tracking
Trajectories
Graph Trajectory similarity (GTS)
Node similarity function (NSF)
Weight adjacency matrix (WAM)