Extraction of urban green space in shadow area from IKONOS image

In this paper the extraction of urban green space under the shadow from an IKONOS image was discussed and a hierarchical classification technique combining textural features and spectral characteristics was presented. Gulou campus of Nanjing University was taken as the test area. Firstly, the spectral characteristics of different features in all bands were analyzed and Difference Vegetation Index (DVI) was employed to extract part green space and eliminate other features meanwhile, such as buildings in non-shadow area, part roads in shadow area. Then, based on the different spectral features, the roads and bushes which are in non-shadow areas were extracted and eliminated hierarchically. Finally, variance of textural feature information was employed in post processing and green spaces in shadow area were extracted. The results indicated that the method is an effective way to extract green space in shadow.