A novel predictor for moving objects

Existing trajectory prediction algorithms mainly employ kinematical models to approximate real world routes and always ignore spatial and temporal distance. In order to overcome the drawbacks of existing trajectory prediction approaches, this paper proposes a novel trajectory prediction algorithm. It works as: (1) mining the interesting regions from trajectory data sets; (2) extracting the trajectory patterns from trajectory data; and (3) predicting the location of moving objects by using the common movement patterns. By comparing this proposed approach to E³TP, the experiments show our approach is an efficient and effective algorithm for trajectory prediction.