基于时空金字塔匹配的轨迹相似度算法

Trajectory Similarity Metric Based on Spatio-temporal Pyramid Matching

轨迹相似性度量是轨迹数据挖掘的基础问题。受设备型号、信号强度和周围环境的影响,轨迹数据具有噪声大、数据量大、采样不均匀等特征,给轨迹相似性度量带来了极大的挑战。因此,提出了基于时空金字塔匹配的轨迹相似度算法,通过在时间和空间维度上对轨迹进行不同粒度的划分,然后利用不同粒度的权重组合来衡量轨迹之间的相似性。该算法能够有效克服轨迹噪声的影响,同时兼顾了轨迹的时间特性和空间特性,并具有较低的计算复杂度。最后,利用真实的信令数据集和人工合成的全球定位系统数据集进行实验,实验结果证明了该算法在准确率和计算复杂度方面都优于目前的主流算法。

The measurement of trajectory similarity is a fundamental problem in trajectory data mining.Trajectory data,affected by devices,signal strengths,and surrounding environments,are characterized by high noise,large volumes,and uniform sampling,which pose significant challenges for measuring trajectory similarity.Therefore,a novel trajectory similarity metric based on spatio-temporal pyramid matching(STPM)is proposed,which divides trajectories into different granularities in the temporal and spatial dimensions and then utilizes weighted combinations of different granularities to measure the similarity between trajectories.The metric effectively overcomes the influence of trajectory noise while considering both the temporal and spatial characteristics of trajectories,and has low computational complexity.Finally,experiments are carried out using real signaling data set and synthetic global positioning system(GPS)data set,and the experimental results prove that the proposed metric is superior to other state-of-the-art similarity metrics in terms of accuracy and computational complexity.