Automatic quality assessment and update of road data in sub-urban areas using aerial images

点击翻译按钮获取中文摘要

Road databases have to be checked regularly to ensure their correctness and completeness, especially in suburban areas due to frequent changes. In the first project period a strategy for the verification of a road database in suburban areas was developed to assess the correctness of the road data. The topic for the second project period is the update of a road database in suburban areas, to additionally ensure the completeness of the database. In suburban areas it is often not possible to extract a road as a whole due to the complex surroundings, for example trees, shadows, and vehicles. In the update phase, we will use the road part extraction strategy developed in the first project period to obtain reliable road parts. Neighbouring road parts are then assembled locally to subgraphs. Context objects are used to explain the gaps between road parts. After assembling the road parts to subgraphs, they must be linked together and connected to the already existing database roads to form a road network. Due to the typical structure of road networks in suburban areas, a global optimisation is impractical here. Therefore, we will pursue a more local approach, setting up junction hypotheses that can connect the subgraphs and verifying them using context objects.