Automated Extraction of Impermeable Land Surface in Central City:A Case Study of Nanjing
SUN Yu,WU Guo-ping,LIU Dong
DOI: https://doi.org/10.3969/j.issn.1000-3177.2013.06.011
2013-01-01
Abstract:Impermeable Land Surface(ILS)area provides an essential indicator of environmental quality and valuable input for planning and management activities.In this paper,we put forward a method which could be used to extract impermeable land surface from multi-spectral remotely sensed image of IKONOS automatically.The method is based on the advantages of pixelbased and object-oriented methods when they are used to extract objects from remote sensing imagery.Firstly,compute NDVI, NDWI,and HSV color model from the original image,and got the thematic map of vegetation,water,and bare land through the threshold segment for NDVI,NDWI and Hue of HSV color model by the algorithm of OTSU;Then,remove shadows in the thematic map of water by using object oriented method;Finally,obtain the thematic map of impermeable land surface by the Boolean operation on the thematic maps of vegetation,water,and bare land.In this paper,we also extracted impermeable land surface of Nanjing's central area,and the high-resolution satellite imagery used in the experiment is multi-spectral imagery scanned by IKONOS sensor.The experimental results show that the extraction of impermeable land surface in city center,using the method proposed in the paper,from IKONOS multi-spectral imagery was good on the whole:the user's accuracy was 93.40%,the producer's accuracy was 95.83%,and the Kappa coefficient was 89.29%.