时空轨迹中隐含的周期性行为模式

将人类时空轨迹中隐含的信息进行可视化并转换为语义语言是公安工作的迫切需要。使用聚类算法将空间轨迹数据离散化,设计等间距抽样算法规整时间序列数据,从而将连续域上的时空轨迹数据表示为离散的二值序列,为使用离散傅里叶变换和循环自相关方法查找时空轨迹中隐含的周期性行为模式做好数据准备;并通过调整采样时间跨度以增强稀疏二值序列的周期性特征。实验表明可以有效提取时空轨迹中的周期性行为模式,而具有季节性和随机变化的行为模式也具有重要参考价值。案件的发生是由人的行为导致的,所以对人类时空轨迹的研究成果也可以应用于案件分析。

Periodic Behavior Patterns Implied in Spatiotemporal Trajectory

Visualizing the implicit information in human spatiotemporal trajectories and converting them into semantic language are the urgent needs for the public security works. Utilizing clustering algorithm to discretize the space trajectory data, and design an equally spaced sampling algorithm to regulate time series data, thus translate spatiotemporal trajectory data in continuous domain into discrete binary sequences, and then make data preparation for finding periodic behavior patterns with the discrete Fourier transform and circular autocorrelation methods, and by adjusting the sampling time span to enhance the periodic characteristics of the sparse binary sequences. Experiments show that can extract the periodic behavior patterns in spatiotemporal trajectories effectively, and the seasonal and random behavior patterns are also useful references. Because the cases are caused by human behaviors, so the researches on human spatiotemporal trajectories can be applied to the case analysis.