出租车轨迹数据的频繁轨迹识别

为识别城市交通中的频繁路径，本文提出了一种出租车轨迹数据的频繁轨迹识别方法。该方法首先对轨迹数据进行轨迹压缩，以降低计算复杂度；然后基于最长公共子序列和动态时间规整算法进行轨迹相似性度量计算，利用计算得到的轨迹间相似度生成距离矩阵；最后将生成的距离矩阵结合HDBSCAN算法进行聚类得到频繁轨迹。选取厦门岛内两个区域进行试验分析，结果表明，该方法能够识别出轨迹数据集中的频繁轨迹，进而得到城市区域之间通行的频繁路径，对道路规划、路径优化与推荐、交通治理等应用提供帮助。

Frequent trajectory recognition of taxi trajectory data

In order to identify the frequent paths in urban traffic, this paper proposes a method of frequent trajectory identification for taxi trajectory data. The method firstly compresses the trajectory data to reduce the computational complexity. Then calculates the trajectory similarity measure based on longest common subsequence and dynamic time warping algorithm, and generates a distance matrix by using the calculated similarity between trajectories. Finally, the generated distance matrix is clustered with HDBSCAN algorithm to get frequent trajectories. Two areas in Xiamen Island are selected for experimental analysis. The results show that the proposed method can identify the frequent trajectories in the trajectory data set and obtain the frequent paths between urban areas, which is helpful for road planning, path optimization and recommendation, traffic management and other applications.