Building roof extraction as data for suitability analysis

Abstract

The geographic potentials of building roofs in urban areas are often under-utilized or sometimes neglected entirely. This is due to the lack of large-scale building roof inventory in the municipality planning offices in African cities. Building rooftops offer great potentials in addressing key urbanization problems such as suitability analysis, disaster management, and others. This paper seeks to demonstrate how remote sensing techniques can be applied to detect and extract building roofs for various suitability analyses (e.g., solar energy and rainwater harvesting) by developing a unique two-in-one technique, which detects and extracts building roof planes and building roof outlines. Using the City of Cape Town, South Africa, as a case study, building rooftops were extracted using the integration of light detection and ranging (LiDAR) data and aerial imagery in an object-based rule-set classification process. This resulted in a building roof inventory, which has crucial attributes of evaluating a building roof for various suitability analyses. The technique developed is well suited for residential and industrial areas where buildings are easily separated from each other. The method and the extracted results were accessed for accuracy and were found to compare well with similar methods already benchmarked.