Extraction of buildings in heterogeneous urban areas: testing the value of increased spectral information of WorldView-2 imagery

Large-scale spatial data that include both topographic and thematic information are needed to support decision-making and regular land management activities at municipal level. Buildings are one of the main feature classes of interest for a municipality, whose ‘correct’ automatic extraction remained a challenging task, even with the advent of Very-High Resolution (VHR) satellite imagery. Using Lisbon as the case study, this work tests the contribution of the increased spectral resolution of the WorldView-2 sensor for building extraction, in an urban planning context. The methodology is based on automated feature extraction and object-based accuracy assessment. Results show that, while the overall accuracy of extraction is relatively high, it has not increased significantly when the newly-available bands were included.