A Fine Track Initiation Algorithm for Group Targets Based on Nearest Neighbor and Graph Theory
Jimin Li, Xingxiu Li, Panlong Wu, Jian Wu, Fanjing Huang, Songtao Li · 2022 International Conference on Cyber-Physical Social Intelligence (ICCSI) · 2022
For the challenge of fine tracking initiation of dense targets within a group, a fine tracking initiation algorithm for group targets based on nearest neighbor and graph theory is proposed. First, a connectivity graph is established based on the distribution structure of the trajectory measurements and the nearest neighbor principle, then, a weighted adjacency matrix is established from the graph information, and finally, a depth-first traversal is performed based on the connectivity graph and the adjacency matrix to determine and optimize the fine starting trajectories of the targets within the group. The simulation verifies that the proposed algorithm improves the correct track initiation rate; reduces the false track initiation rate; significantly reduces the correlation of multiple point tracks between adjacent moments; and improves the overall accuracy of the initiated tracks compared with the modified logic method, DBSCAN-based and modified Hough transform multi-trail initiation algorithms.