Trajectory simplification plays an important role in trajectory data storage,transmission,temporal-spatial analysis and visualization.Traditional simplification methods,such as Douglas-Peucker algorithm,concern the geometric information while ignore the temporal information,which may result in loss of implied mobility features in the original trajectory.Aiming at minimize speed error in the trajectory simplification transformation,this paper presents a new method based on hierarchical clustering and regional ization operations.First,the line segments of the original trajectory are clustered at different levels based on the similarity of speed measure.With the support of the hierarchical clusters,the original trajectory is then divided into a series of segments.For each segment,the maximum synchronized Euclidean distance from the points to the segment line connecting two end points is no larger than the predefined threshold value.Finally,the simplified results is outputted by organizing the end points of each trajectory segments.Real life data was used to verity the effectiveness of the proposed method,and results of comparing with other existing methods showed that our method performs better in speed preserving.%轨迹线数据实施化简处理对于缓解数据存储、传输压力以及后期的分析可视化效率具有重要意义.常规方法(如Douglas-Peucker算法)主要考虑线目标的几何形态结构,直接应用到轨迹线化简中容易丢失移动物体的运动状态特征.本研究从保持轨迹线隐含速度特征出发,提出了一种基于移动速度相似性原则的轨迹线层次化剖分与分区化简处理方法.首先,以相邻轨迹点构成的直线段为基本单元,在拓扑连接关系约束下基于速度指标对轨迹直线段进行层次化聚类,并将聚类结果组织为层次树结构;然后,以建立的层次树结构为约束条件对原始轨迹线实施分区处理,使得同一区域内轨迹线片段的中间点距首尾基准线的最大时间同步偏移距离小于设定的阈值;最后,依次连接各分区轨迹线片段首尾点导出化简结果.采用真实的车辆轨迹线作为试验数据,通过与其他多种方法进行对比分析验证了本文提出方法的有效性.
展开▼
