Reconstruction of the trajectories of moving objects using context-based dynamic time warping similarity measure method

With the increasing growth of positioning technologies and the use of navigation systems, a large volume of moving point object data, such as people, cars, ships, and animals, is available. However, the lack of integrity and incompleteness of these data for systemic, human, and environmental reasons challenges the analysis of trajectories and their effective application in various fields. Therefore, the reconstruction of missing data plays an important role in maximizing the capacity of movement data, particularly in navigation and track tracking. In this study, using the similarity measurement of trajectories approach, trajectories containing gaps are reconstructed. In this regard, the context-based dynamic time warping (CDTW) method, along with speed and direction movement parameters, are used to measure the similarity and reconstruct the trajectories of vessels in two regions of the Atlantic and Pacific Oceans. Two mechanisms, a constant number of trajectories and a specified threshold, are considered for reconstruction. The results show that using a constant number of trajectories in comparison with the specified threshold reduces the root mean square error (RMSE) and mean absolute error (MAE) from 1.5 and 1.4 to 0.5 and 0.4, respectively. In addition, increasing the length of the trajectories improves the RMSE and MAE values from 0.5 to 0.1 in the case of a constant number of trajectories and 1.5 to 0.3 in the case of the specified threshold.