For vehicles equipped with GPS devices, their recorded trajectories are often erroneous or at least inaccurate. Thus, trajectory enhancement is required for applications that need precise vehicle locations. Map matching is a widely used technique to enhance trajectories with the help of maps. However, the results by map-matching are usually not reliable due to incomplete or inaccurate map information and inadequate sampling rates of trajectories. In this thesis, we propose a visual analytics system to help domain experts analyze the map-matching results, correct wrong results, and detect trajectory error patterns. Our system consists of three major components: a global viewer summarizing the map-matching results, an area viewer presenting the road information and the associated trajecto...[ Read more ]

For vehicles equipped with GPS devices, their recorded trajectories are often erroneous or at least inaccurate. Thus, trajectory enhancement is required for applications that need precise vehicle locations. Map matching is a widely used technique to enhance trajectories with the help of maps. However, the results by map-matching are usually not reliable due to incomplete or inaccurate map information and inadequate sampling rates of trajectories. In this thesis, we propose a visual analytics system to help domain experts analyze the map-matching results, correct wrong results, and detect trajectory error patterns. Our system consists of three major components: a global viewer summarizing the map-matching results, an area viewer presenting the road information and the associated trajectory summary, and a trajectory viewer showing the original trajectory details. We develop two novel encoding schemes to present the statistics and clusters of trajectories over maps, which can help users identify different factors leading to wrong matching results. The case studies with the real trajectories of thousands of vehicles have demonstrated the effectiveness and usefulness of our system.