基于模式匹配的目标轨迹预测方法

Prediction Method Based on Target Trajectory Pattern Matching

摘要: 本文在目标轨迹预测中采用了数据挖掘的方法，提出了一个具体的基于移动模式匹配的目标轨迹预测算法。该方法通过不断挖掘历史移动轨迹来构造前缀共享树的方法挖掘出频繁移动模式，之后通过模式匹配预测出目标的移动轨迹。仿真结果表明该算法的时间消耗和空间消耗较小，同时具有很高的预测准确性。

Abstract: A specific target trajectory prediction algorithm based on mobile pattern matching is proposed, which is called PS-Tree algorithm. In this algorithm, historical data of targets’ movements generated in monitored region are used for pattern mining, coordinate series of targets are converted to region trajectory series, and frequent moving modes are figured out by building up PS-tree. The algorithm matches current trajectories with the ones in pattern library to forecast the movements of the target when a target moves into monitored region. The simulation results prove that PS-Tree algorithm has low time and space consumption but high prediction accuracy.