October 2012
·
85 Reads
·
25 Citations
We discuss methods to visualize large amounts of object movements described with so called multivariate trajectories, which are lists of records with multiple attribute values about the state of the object. In this chapter we focus on vessel traffic as one of the examples of this kind of data. The purpose of our visualizations is to reveal what has happened over a period of time. For vessel traffic, this is beneficial for surveillance operators and analysts, since current visualizations do not give an overview of normal behavior, which is needed to find abnormally behaving ships that can be a potential threat. Our approach is inspired by the technique of kernel density estimation and smooths trajectories to obtain an overview picture with a distribution of trajectories: a density map. Using knowledge about the attributes in the data, the user can adapt these pictures by setting parameters, filters, and expressions as means for rapid prototyping, required for quickly finding other types of behavior with our visualization approach. Furthermore, density maps are computationally expensive, which we address by implementing our tools on graphics hardware. We describe different variations of our techniques and illustrate them with real-world vessel traffic data. © 2013 Springer Science+Business Media New York. All rights are reserved.