面向安防系统的高效用语义轨迹模式挖掘

High-Utility Semantic Trajectory Pattern Mining for Security System

在安防系统中,将大量目标轨迹先转化为语义轨迹后再进行频繁模式挖掘,有助于分析目标行为模式、识别危险源及增强安防系统内部防控。针对现有频繁模式挖掘方法未考虑目标停留点的效用差异问题,提出一种高效用语义轨迹模式挖掘算法。综合停留点兴趣度、目标停留时间以及目标语义轨迹支持度这3个参数定义语义轨迹效用值,采用蚁群算法挖掘高效用语义轨迹模式。利用精英蚂蚁策略改进蚂蚁种群的迭代方式,通过轮盘赌选择法优化蚂蚁对于下一个节点的选择策略,运用无效用编码向量剪枝策略提高算法执行效率。在Chess、Mushroom、Foodmart、Retail等4个公开数据集以及某安防系统的RFID定位数据集上的实验结果表明,相比于HUIM-ACS算法,该算法挖掘的高效用语义轨迹模式数量增加了10%~15%,运行时间减少了7%~12%。

In security systems,transforming a large number of collected target trajectories into semantic trajectories and mining their frequency patterns are helpful in analyzing target behavior patterns,identifying hazard sources,and enhancing internal prevention and control of security systems.Existing frequent-pattern mining methods do not consider the utility difference in stay-point values.To address this issue,this study proposes a high-utility semantic trajectory pattern mining method.The concept of semantic trajectory utility value is defined by integrating three parameters:the interest of the stay point,stay time of the target at stay point,and the support of target semantic trajectory.To achieve this,an ant colony algorithm is used to mine high-utility semantic trajectory patterns.The algorithm involves the elite ant strategy to improve the iterative method of ant population and the strategy of the ant selecting the next node through the roulette selection method.Next,the invalid coding vector-pruning strategy is used to improve the execution efficiency of the algorithm.The proposed algorithm is tested on four public datasets,namely Chess,Mushroom,Foodmart,and Retail,as well as the Radio Frequency IDentification(RFID)location dataset of a certain security system.The experimental results show that the proposed algorithm increases the number of high-utility semantic trajectory patterns by 10%-15%and reduces the running time by 7%-12%compared with the Ant Colony Optimization(ACO)-based approach to mine high-utility itemsets(HUIM-ACS).