Retrieval of urban residential areas from the fused image of SAR and TM data based on HIS transformation

This paper aims at improving the retrieval accuracy of urban residential areas in and region where barrens are distributed widely by fusing multi-spectral image and SAR image based on HIS transformation. Unlike traditional supervised classification, 2-class classification is used to obtain urban residential areas and non-residential areas. Comparison between this method and traditional supervised maximum likelihood classification (MLC) method is performed. It is found that the retrieval of urban residential areas from the fused image is more accurate than that from TM image by using MLC, with accuracy of 84.21% and 71.79% respectively. The result validates the efficiency of our method.