Research On Simplification And Semantics Enhancement Of Spatio-Temporal GPS Trajectory From Travel Survey

Today's outstanding urban traffic problems indicate that the capacity of transport system has been overpowered by daily travel demands of the urban dwellers. This and other related social issues become severe restrictions to sustainable urban development in China. By analyzing personal travel data collected through transportation survey, useful information about the characteristics and distribution of public demands on transportation services and other urban functions can be derived and used to support urban planning and management. Positioning technologies such as GPS have made it feasible to collect accurate and detailed personal travel data with finest time granularities; however, the data also post great technical challenges in terms of complex storage structure, large volumes, and semantic annotation before becoming useful. All these issues require special methods for data processing, analysis, interpretation, and application. It is the purpose of this thesis, therefore, to develop simplification and knowledge mining techniques for traffic information extraction from GPS travel trajectory data.Based on characteristics of urban travel behaviors and GPS-based travel survey, this paper aims to address the spatiotemporal simplification and semantic enhancement of personal travel trajectories through the following three data processing phases. The structure and functions of the GPS data processing platform and the visualization method of trajectory are designed in the following work. All these work will be the basis of further data mining and knowledge discovery.First, data cleaning is conducted by eliminating random noise and abnormal trajectory points, followed by data simplification based on speed and heading change characteristics.Second, sub-trajectories are identified by segmenting raw trajectories and reconstruction with key points (i.e. Begin Point, End Point, Stops and Moves), which are extracted via a computing algorithm considering both spatial and temporal characteristics.Finally, a coordinate to semantic conversion is performed with knowledge from geographic domain and application domain through spatial join for the reconstructed trajectories.