Effective Trajectory Similarity Measure for Moving Objects in Real-World Scene

摘要: Trajectories of moving objects provide fruitful information for analyzing activities the objects; therefore, numerous researches have tried to obtain semantic from trajectories by using clustering algorithms. In order cluster trajectories, similarity measure should be defined first. Most existing methods utilized dynamic programming (DP) based measures cope with different lengths trajectories. However, DP do not enough discriminative power properly real-world environment. this paper, an effective trajectory is proposed, and proposed on geographic similarities which a same scale. Therefore, importance can easily controlled weighted sum two similarities. Through experiments challenging dataset, was proved better than method.