摘要
随着卫星定位传感器的普及应用,形成了海量移动对象的轨迹数据。轨迹数据含有丰富的时空特征信息,通过对相关数据聚类处理,可以挖掘出移动对象的活动场景、位置等属性信息。通过借鉴神经成像学领域中的QuickBundles算法,介绍算法原理和实现,并基于此算法实现了一种轨迹聚类方法,通过使用实际GPS数据对此方法进行验证,从对聚类结果的可视化展示表明,本方法能够有效挖掘原始数据,完成对轨迹数据的聚类。
With the popularization of satellite positioning sensors,a large number of trajectory data of moving objects have been formed.The trajectory data contains rich spatio-temporal feature information.By clustering the relevant data,the attribute information such as activity scene and location of moving objects can be mined.By referring to the QuickBundles algorithm in the field of neuroimaging,the principle and implementation of the algorithm are introduced,and a trajectory clustering method is implemented based on this algorithm.This method is verified by using the actual GPS data.Through the visual display of the clustering results,this method can effectively mine the original data and complete the clustering of trajectory data.
作者
刘峰
Liu Feng(Jiangsu Automation Research Institute,Lianyungang,Jiangsu 222002,China)
出处
《计算机时代》
2022年第4期43-46,共4页
Computer Era
关键词
轨迹数据挖掘
聚类
可视化
应用
trajectory data mining
clustering
visualization
application