ABSTRACT
Mapping urban areas at the regional and global scales has been used in ecology, environment, sociology, and other subjects. Recently, it has become increasingly popular to extract urban areas from night-time light remote-sensing data. In this article, we reported an alternative method to extract information of urban areas from VIIRS Day/Night Band (DNB) and MODIS normalized differential vegetation index (NDVI) data based on the adaptive mutation particle swarm optimization (AMPSO) algorithm and the Support Vector Machine (SVM) classification algorithm. This method was validated using the urban areas of nine Chinese cities classified from Landsat Enhanced Thematic Mapper (ETM+) images by object-oriented image classification technology. We demonstrated that this new method for urban area extraction had a good classification coherency with the Landsat8 OLI result. In addition, it is more robust than other classification methods, and can be used to characterize the inter-urban texture as well.
