Answers are grounded in this paper and cite their sources.
Because of the satellite orbit height and solar elevation angle, remote sensing imagery of urban areas with fine spatial resolution inevitably contains shadows cast by buildings, which thus decreases the accuracy of target recognition and information extraction. This paper proposes a new normalized-difference shadow detection index based on the saturation and near-infrared information of imagery to detect building shadows. The method was validated using typical images of urban areas from the WorldView-2, TripleSat-2, and RapidEye satellites and compared with the C3 component, multispectral band shadow detection, principal component analysis shadow detection (PCAS), normalized-difference umbra index, and GeoEye-1 shadow detection index methods. The results indicated that the proposed method effectively distinguishes building shadows from other dark ground objects and provides higher detection accuracy. The recall rate, precision, and F1 score exceeded 85%, 80%, and 85%, respectively, and their average values for four representative images were 91%, 89%, and 90%, respectively. Our method provides greater usability and better performance in the detection of building shadows in remote sensing imagery with fine spatial resolution.
No takes yet. Share an insight, caveat, or question.
Fang et al. (2019) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: