Building Visual Summaries of Clusters of Trajectories

A big problem in analysis of movement data (trajectories of
moving agents) is how to visualize clusters of trajectories so as to
have an overview of all clusters and to be able to compare
different clusters. Clusters of trajectories are not disjoint in space
but intersect and overlap. Hence, they need to be shown separately
in multiple small displays. Such displays can be legible only if the
clusters are presented in a schematized form. We suggest a
method for building schematic representations of clusters of
trajectories. The method transforms trajectories into aggregate
flows between areas. The areas are automatically defined on the
basis of the spatial distribution of the characteristic points from
the trajectories. We use numeric measures of the abstraction
quality and techniques for controlling the quality.