Application of an Adaptive and Directed Kernel Density Estimation (AD-KDE) for the Visual Analysis of Traffic Data
GeoViz: Linking Geovisualization with Spatial Analysis and Modeling, 10-11 March 2011, Hamburg, Germany
Jukka M. Krisp, Stefan Peters, and Masria Mustafa
“Remote sensing and tracking technologies (such as tracking mobile phones and in vehicle navigation systems) have enabled us to store the position of individual cars over time. A number of previous studies have investigated methods to display the density of cars on a road network. These density maps show density of vehicles moving on the street as an isopleths map within a city or region at a single instance. Often these investigations use the well known kernel density estimations (KDE) to display static densities (and traffic hot-spots) representing one point in time. This “classic” method may reveal the traffic hotspots, but it does not recognize the movement trends within dynamic point data representing the individual cars. Therefore we investigate how to display the dynamic component of point densities and the density changes. We take two points in time (set1 and set2) representing the individual vehicles positions and apply an adaptive directed kernel density estimation (AD-KDE), which recognizes the underlying dynamics of the individual points.”
- Read the paper [PDF]
