Dubai is a modern city which is growing at a fast pace, constantly changing and developing. The application presented in this paper is part of the “Smart Application for Feature extraction & 3D modelling using high resolution satellite Imagery (SAFIY)” project, whose main aim is to develop satellite data applications to help the government in Dubai to get up-to-date information on features such as water bodies, vegetation areas, buildings and roads. In this paper we present an application to automatically update the road network map for the city of Dubai using very high resolution satellite imagery, Deimos-2 and DubaiSat-2 at 0.75 cm resolution. Deimos-2 and DubaiSat-2 have been recently included as third party contributors to the Copernicus constellation. The algorithms implemented use unsupervised and supervised classification techniques to extract the road pixels. Special road objects such as roundabouts or bridges are also detected using algorithms based on feature descriptors, designed to match the specific characteristics of those objects. Finally, the system identifies the changes in roads and objects from the current road network map in order to update the map The city of Dubai is 4,114 km~2 and has many different types of neighbourhoods from suburbs to very densely populated areas. Overall, we propose a novel end to end system that automatically updates a road network map and allows a user to manually correct them so that the final resulting map is usable.
展开▼
