Nowadays, the main methods for obtaining land cover classification information in remote sensing data are still visual interpretation and artificial identification. These methods can obtain more accurate classification results, but their efficiency is too low to meet the increasing application demands. By taking Guangdong province as an example, an automatic land cover classification information extracting and dynamic monitoring method based on remote sensing images is studied. The software for extracting land use classification and dynamic information is developed. Automatic or semi-automatic land use classification is realized. The result can meet the needs of the second national land survey and provide the necessary guidance for land use planning.