The wetland classification of medium-resolution remote sensing images is a difficult problem because the mixed pixel phenomenon is serious. In this study, visual interpretation was used to extract water. Then the traditional and improved linear spectral mixture analysis models were selected to extract other wetland types. Although errors and confusion exist, this method shows satisfying results with an overall accuracy of 86.2% and a KAPPA coefficient of 0.79. The method in this paper can achieve a more precise information extraction of the wetland and could provide a reference of wetland extraction for other regions.
展开▼