Track-to-Track Association Algorithm with Spatio-Temporal Features
Hao Lang, Siwei Yu, Wei Zhong Ren, Chaoyi Han, Xiaoyu Hu, Zhiduo Ji · 2024
Track-to-track association aims to solve maneuvering target attitude estimation during multi-source information track fusion processing. However, high accuracy is not achieved by the typical track-to-track association algorithms due to the impacts of random noises, omissions, and system bias. To address this issue, we propose a spatio-temporal track association method that compares and analyzes the target to be associated with other nearby targets by using the relative position relationship as the criterion of association judgment. First, this method uses the topological structure information between targets to complete track association. Second, the modified Hausdorff distance serves as the metric to introduce the historical state of the track. Then, The topological distance and the modified Hausdorff distance are linearly weighted to create the association cost matrix. Finally, trajectory correlation is completed by applying a linear assignment algorithm. To verify the accuracy and reliability of the algorithm, we set up a simulation experiment scenario and conducted experiments. Comparing GNN and NN algorithms under two scenarios, the accuracy increased by 17.25%, 11.28%, and 46.22%, 13.02% respectively. Our experimental results demonstrate the method performs well in different scenarios.