Advanced techniques in trajectory data analysis for anomaly detection and map construction

With a large amount of trajectory data generated every day, there is a high demand for developing advanced techniques to discover the underlying information instead of dull and heavy manual work. This thesis focuses on the anomaly detection and map construction from GPS data. Anomaly detection aims...

Descripción completa

Detalles Bibliográficos
Autor: Yuejun, Guo
Tipo de recurso: tesis doctoral
Estado:Versión publicada
Fecha de publicación:2020
País:España
Institución:CBUC, CESCA
Repositorio:TDR. Tesis Doctorales en Red
OAI Identifier:oai:www.tdx.cat:10803/673055
Acceso en línea:http://hdl.handle.net/10803/673055
Access Level:acceso abierto
Palabra clave:Dades GPS
Datos GPS
GPS data
Mapes
Mapas
Maps
Cartografia
Cartography
Detecció d'anomalies
Detección de anomalías
Anomaly detection
Anàlisi de dades de trajectòries
Análisis de datos de trayectoria
Trajectories data analysis
68
91
Descripción
Sumario:With a large amount of trajectory data generated every day, there is a high demand for developing advanced techniques to discover the underlying information instead of dull and heavy manual work. This thesis focuses on the anomaly detection and map construction from GPS data. Anomaly detection aims to identify trajectories that do not follow common behaviors, and map construction deals with a set of trajectory data to generate a route graph that represents the main movement paths hidden in data. For online anomaly detection, we study the well-known Sequential Hausdorff Nearest-Neighbor Conformal Anomaly Detector (SHNN-CAD) approach, and propose an enhanced version called SHNN-CAD•. For map construction, we present a new, fast and robust three-step framework. Considering the storage limitation and computational cost dealing with large-scale data, we propose a split-and­merge strategy. Besides, we utilize the edge weight to visualize the map and remove the wrong edges.