Urban land cover classification based on WorldView-2 image data

Cities hahave a complex construction and easily affected by human activities, so they needsto be surveyed and analyzed timely. WorldView-2 high resolution remote sensing image makes it possible to study the urban land cover classification by its abundant space geometric features and spectral information. This paper would have aimed at given urban land cover types to choose suitable segmentation scales and classification features through object oriented multi-scale segmentation and classification method based on WorldView-2 image data, and extracted the urban land cover types progressively according to reasonable order. Then it raised the NDWI and NDVI which appropriated to extract water and vegetation on WorldView-2 image, and grouped the objects' features after segmentation to extract the roads and buildings hierarchically. The results of accuracy assessment indicated that using this method to study the urban land cover classification based on WorldView-2 image received an ideal effect.

Перевод пока недоступен