Impervious Surface Information Extraction Using an Improved Object-Oriented Method

Impervious surface, where water cannot infiltrate the soil, including roads, driveways, sidewalks, parking lots, rooftops, and so on, has been recognized as an important indicator in urban environment. However, accurate extraction of impervious surface information from imagery is still a challenge in remote sensing society. This paper explores extraction of impervious surface information with QuickBird imagery based on improved object-oriented. Firstly, MNF (minimum noise fraction) was used to achieve end members. Secondly, vegetation fraction image was produced by linear spectral mixture analysis. Thirdly, three urban land-use classes were developed based on object-oriented method. Results showed that dark impervious objects from shadows cast by tall building and tree canopy had been separated from water.