关键词: TM图像, 土地利用/覆被, 分层信息提取, 归一化植被指数, 县域

Abstract:

It is important to achieve the qualitative and quantitative information of land use/cover in a county (or county-level city) of China with higher precision, which is helpful to enhance eco-environment protection and sustainable development of rural economy. Presently, remote sensing images of the medium and high resolution are mostly used to monitor the changes of land use/cover in a county. To make better use of remote sensing technology in monitoring land use change, it is necessary to improve the automatization level of information extraction from the remote sensing images and meet the precision of change monitoring synchronically. Hierarchical information extraction is an effective method for information extraction of land use/cover in a county from the remote sensing images. Based on the information of each ground object, the image is decomposed layer upon layer according to certain principles. The method functions well in the classification precision over "the same object with different spectra" and "different objects with the same spectrum" because the environment of information extraction is comparatively pure. Aided by the TM image in Xinyi city of Jiangsu province obtained in the winter of 2003, the hierarchical method is used to extract the information of land use/cover. On the basis of such processes of original images as geometric correction, image registration, image clip and image enhancement, first of all, the image was classified using the maximum likelihood classifier and the unused land with the least inaccurate probability was extracted from the classification result observed. Then, the scope of the water was extracted using spectral analysis method, the urban construction land and village by combining supervised classification method with visual interpretation method, and the woodland using Normalized Difference Vegetation Index (NDVI). Finally, the cropland was extracted. By comparing extracted result of land use/cover with the land use map in the same period, the area accuracy of the land use classification for the entire Xinyi city reaches 96.17%, and the space accuracy reaches 88.38%. These indicate that the hierarchical method applied to information extraction of remote sensing images is feasible.

Key words: TM image, land use/land cover, hierarchical information extraction, Normalized Difference Vegetation Index, county