Fast extraction of roads for emergencies with segmentation of satellite imagery

Among the problems that national and international organizations often face in rescue operations following earthquakes there is the knowledge of the roads that allow aid to reach its destination. During the Haiti earthquake on 2010-01-12 and the Nepal earthquake on 2015-04-25 proved to be very useful the work of volunteers creating maps of the roads on site, useful in case of emergency. Though away from the site of interest, with such crowdsourcingactions,volunteers can contribute to the efforts of mapping inaccessible or little known regions by tracing images from aerial photographs or satellite images, in some cases made available free of charge. The objective of this research, still in progress, is to experience for obtaining very quickly the road network, by extracting it automatically or semi-automatically from satellite data with the Object Based Image Analysis, with a segmentation based on Mathematical Morphology.