A Morphological Feature-Oriented Algorithm for Extracting Impervious Surface Areas Obscured by Vegetation in Collaboration with OSM Road Networks in Urban Areas

Remote sensing is the primary way to extract impervious surface areas (ISAs). However, obstruction of vegetation a long-standing challenge that prevents accurate extraction urban ISAs. Currently, there are no general and systematic methods solve problem. In this paper, we present morphological feature-oriented algorithm, which can make use OSM road network information remove obscuring effects when ISAs extracted. Very high resolution (VHR) images Wuhan, China, were used in experiments verify effectiveness proposed algorithm. Experimental results show algorithm improve accuracy completeness ISA by our previous deep learning-based overall (OA) 86.64%. The feasible vegetation-obscured effectively precisely.