中国科学院软件研究所机构知识库

城市中海量人群的移动数据可视化对解决当代城市面临的一些问题（如交通优化、商业选址等）起到重要的作用。

在本工作中，我们基于一组覆盖数百万城市居民的大型移动定位数据集展开研究，该数据的显著特征为单个用户轨迹具有时间稀疏的特性。采用稀疏轨迹分析、可视化城市中的人群移动面临着诸多挑战。现有文献中的轨迹数据分析和可视化研究主要针对记录间隔为秒级或分钟级的稠密轨迹数据，但尚不支持记录间隔为数小时的稀疏轨迹。本工作从数据集中发现了轨迹的长尾稀疏性，进而提出了针对长尾稀疏轨迹的分析和可视化框架。在分析方面，本工作应用了一个有效的方法来从长尾稀疏轨迹中提取移动，基于DBSCAN对局部空间内的移动进行聚合进而形成一个向量场，并使用统计学中的方法从移动数据中发现了特征和异常模式；在可视化中，我们借鉴了风图的可视化方法来动态地展示人群的移动，并基于确定性纤维束跟踪算法设计了从局部移动向量生成全局移动流的方法。基于上述方法，我们实现了“UrbanMotion”可视分析系统，该系统基于原始的风图技术进行了拓展，能够展示多方向的人群流动和基于热力图的人群分布情况。

我们围绕通勤分析、事件检测、广告投放三个应用领域进行了案例研究，并收集了相关专家的反馈意见。初步实验与调研结果证明，我们的系统对于辅助用户完成相关的分析任务具有实际的效果与应用价值。

Visualizing massive scale human movement in cities plays an important role in solving many of the problems that modern cities face (e.g., traffic optimization, business site configuration).

 In this work, we study a big mobile location dataset that covers millions of city residents but is temporally sparse on the trajectory of the individual user. Mapping the sparse trajectories to illustrate the population movement poses several challenges from both the analysis and the visualization perspectives. Existing literature focuses on the analysis and visualization of the trajectory with a recording interval as long as seconds or minutes; yet does not support sparse trajectory with a recording interval of several hours. This work finds the long-tail sparsity of trajectories from our dataset and proposes a framework to analyze and visualize sparse trajectories. On the analysis side, this work applies an effective method to extract movement from the long-tailed sparse trajectories and proposes a clustering method based on DBSCAN to aggregate local movement into the vector field. Both characteristic and anomalous patterns are discovered from movement data using statistical methods. On the visualization side, this work introduces the wind map to visualize population movement dynamically, and we also design a new method based on Deterministic Tractography to generate global movement flows from local movement vectors. Based on the above methods, we developed a visual analysis system called UrbanMotion. The system extends an original wind map design by supporting the display of multi-directional population flows and heatmap-based visualization of population distribution.

We conducted three case studies and collected the expert feedback in the application domains of commuting analysis, event detection, and advertising. The result demonstrates the effectiveness and significance of our system in helping to complete the key analytics tasks for urban users.