A feature based method for trajectory dataset segmentation and profiling
A feature based method for trajectory dataset segmentation and profiling
复制标题
一种基于特征的轨迹数据集分割和分析方法
DOI：
10.1007/s11280-016-0396-y
复制
发表时间：
2017
影响因子：
3.7
通讯作者：
Zhao Lei
中科院分区：
文献类型：
--
作者：
Jiang Wei;Zhu Jie;Xu Jiajie;Li Zhixu;Zhao Pengpeng;Zhao Lei
位置获取和移动的计算技术的普及产生了海量的空间轨迹数据，这给海量数据的管理和分析带来了巨大的挑战。在本文中，我们专注于子轨迹数据集的轮廓问题，并旨在从原始轨迹提取代表性的子轨迹作为一个子集，称为轮廓，它可以最好地描述整个数据集。这个问题是非常具有挑战性的主题，找到最具代表性的子轨迹集权衡的大小和质量的配置文件。为了解决这个问题，我们从密度，速度和方向流方面对轨迹数据集的特征进行建模。同时，我们提出了我们的两个步骤的方法来选择代表性的轨迹的基础上的特征模型。首先，一种新的轨迹分割算法应用于一个原始的轨迹，以确定其特征代表性的代表性段，并自动估计段的数量和段的边界。然后，一个子轨迹分析方法进行产生的最具代表性的子轨迹的数据集，基于局部启发式进化策略。我们使用北京和上海超过12，000辆出租车生成的两个真实轨迹数据集，基于大量实验评估我们的方法。结果证明了我们的方法在不同应用中的效率和有效性。
The pervasiveness of location-acquisition and mobile computing techniques has generated massive spatial trajectory data, which has brought great challenges to the management and analysis of such a big data. In this paper, we focus on the sub-trajectory dataset profiling problem, and aim to extract the representative sub-trajectories from the raw trajectory as a subset, calledprofile, which can best describe the whole dataset. This problem is very challenging subject to finding the most representative sub-trajectories set by trading off the size and quality of the profile. To tackle this problem, we model the features of the trajectory dataset from the aspects of density, speed and the direction flow. Meanwhile we present our two-step method to select the representative trajectories based on the feature model. First, a novel trajectory segmentation algorithm is applied on a raw trajectory to identify the representative segments concerning their feature representativeness and automatically estimate the number of segments and the segment borders. Then, a sub-trajectory profiling method is performed to yield the most representative sub-trajectories in the dataset, based on a local heuristic evolution strategy. We evaluate our method based on extensive experiments by using two real-world trajectory datasets generated by over 12,000 taxicabs in Beijing and Shanghai. The results demonstrate the efficiency and effectiveness of our methods in different applications.