Abstract: The developed change detection method, by utilizing the non-local filtered SAR intensity RC composition, is sensitive in the extraction of small patches and linear features. Consequently, it will indicate a better performance in practical applications, in particular this method is not constrained by additional requirements, e.g. the spatiotemporal baseline. In this study, taking the intensity RC composite imagery as the data-source, a semi-automatic change detection method is proposed by utilizing color-transformed features. In order to realize the thematic updating of urban-area land, the corresponding data procedures include four primary steps, they are color space transformation, training-sample selection, supervised classification based image segmentation, and change region extraction. Taking Hexi New Town and Jiangbei Developing District (Nanjing) as example, the checking probability is better than 88% with the optimum parameter setting (3 features along with 10 training-sample categories).

Key words: urban change detection, SAR intensity RC composition, color space transformation, COSMO-SkyMed, construction land-use