基于路网探测的轨迹模式挖掘

Trajectory Pattern Mining Based on Road Network Detection

  • 摘要: 针对现有的移动受限轨迹离散化方法效率低、不直观、易丢失移动模式等问题,提出了一种先进行路网结构探测,再基于道路匹配对轨迹进行离散化的方法.算法首先基于数学形态学理论从轨迹中提取出路网结构,然后将轨迹点匹配到路网中的网格中,以网格序列来表示连续的轨迹,最后使用最大频繁序列模式挖掘方法从中挖掘出轨迹模式.实验结果表明,该算法能够快速有效地对轨迹进行离散化,且能比其它算法挖掘出更多更细致的轨迹模式.

     

    Abstract: The existing discretization methods for moving-constrained trajectory have some defects such as inefficiency,non-intuitiveness,and susceptibility to loss of movement patterns. A novel discretization method that is based on road network construction and road matching is proposed. First,the road network from trajectories based on mathematical morphology theory is extracted. Next,it matches the trajectories to the grids on roads and changes the trajectories into grid sequences. Finally,the trajectory patterns using the maximal frequency sequence pattern mining method are detected. Experimental results show that this algorithm can quickly and efficiently discretize trajectories and detect more detailed trajectory patterns than the other algorithms.

     

/

返回文章
返回