Skip to Main Content
Based on characteristics of clear geometry features in high-spatial resolution remote sensing image, the paper presents a study method of land use/cover by object-oriented classification. The object-oriented classification method overcomes salt and pepper phenomena of conventional classification method by using feature object as basic processing units, which are generated from image segmentation. In this process, we consider spectral and shape as two basic factors and also emphasize texture information of surface features. Object-oriented remote sensing image classification method is based on the cognitive model of remote sensing information extraction, which can achieve multi-scale analysis of spatial, meet different scales requirement of surface feature extraction of information, and integrate multi-source data of classification, so as to make classification results more convinced.