Trafficability analysis after flooding in urban areas using probabilistic graphical models

In this paper, a probabilistic graphical model is presented, which assesses roads concerning their trafficability after flooding. Graphical models are used to combine simulation and observation improving the assessment of roads. The simulation of flooded roads can be established by means of Digital Elevation Models (DEM). The observations are derived from remote sensing images. The graphical model build a statistical framework which combines the images and DEM. In this paper, the results of a pixel-based Bayesian Network are presented which show the benefit of the complementing input information. Furthermore, an undirected graphical model is presented, which includes the topology of neighboring pixels to obtain more robust results.