基于分组地图匹配的异常轨迹检测

Anomaly trajectory detection based on grouping map matching

为提高轨迹数据的质量,提出一种轨迹点的多尺度综合映射算法。采用分组匹配的思想,依据每组特性剪枝无效的轨迹点,加快映射轨迹点的连接速度,快速得到路网轨迹。在此基础上,为有效进行异常轨迹的检测,提出一种基于分组地图匹配的异常轨迹检测算法,研究字符串相似性搜索,构建相似轨迹序列寻找代表轨迹,基于交叉点行驶状况对轨迹点进行约减,减少轨迹间的计算量,计算轨迹的异常阈值,判定异常轨迹。在真实的出租车数据集上进行实验,验证了提出方法的执行效率和准确性。

To improve the quality of trajectory data,a multi-scale comprehensive mapping algorithm of trajectory points was proposed.The idea of grouping matching was adopted,invalid trajectory points were pruned according to the characteristics of each group.The connection speed of mapped trajectory points was accelerated,and the road network trajectory was quickly obtained.An algorithm of anomaly trajectory detection based on grouping map matching was proposed,which effectively detected the anomaly trajectory.By researching string similarity search,the similar trajectory sequence was constructed to find the represen-tative trajectory.Some invalid track points were pruned by analyzing the driving conditions at the intersection,the amount of computation between trajectories was reduced.The abnormal threshold value of each track was calculated,which was used to determine the abnormal track.The efficiency and accuracy of the two methods are verified by experiments on real taxi data sets.