Abstract: Based on RS and GIS, the authors build the interpretation mark of land use type with object-oriented classification technology in accordance with the actual situation of Shenyang, to split images with multi-scale segmentation. A classification rule is established with decision tree method and nearest-neighbor method to extract the regional land use information and get the land use/land cover data of two periods. The dynamic change of land use and its driving force in the study area is then analyzed. In the abstraction of surface landscape information, the object-oriented classification fully uses the ground-object spectral information and texture features to overcome mixing phenomenon to the maximum extent due to similarity of different ground-object spectral information and difference of the same spectral information.