期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2008
卷号：XXXVII Part B7
页码：671-672
出版社：Copernicus Publications
摘要：High resolution satellite images such as IKONOS and Quick Bird data provide detail information of land cover at a certain point of time. The small objects in an image represent specific conditions of the land cover and the land use on the area. In this study, a semi automated algorithm and manual assisted algorithm as developed to generate land use map from IKONOS pan-sharpen image with a rule that understand land use from land cover objects, and re-construct land use categories from small land cover objects which were generated by image segmentation processing method. The average size of the image segmentation was changed from 1,000 pixels to 50 pixels, and the homogeneity of the image object was examined by visual interpretation. By some experimental operations, the variance level to identify an homogeneous land cover was define as 300-700 depend on the land cover type for the IKONOS image, and about 15,000 objects from the test site image, which consists of 1 million pixels, were extracted as land cover objects(segmentation). The land cover categories such as "deciduous forest", "ever green forest", "paddy filed" and some "orchard" with the variance level of 300 with 500 pixel size of image objects show homogeneous land cover type. And the land use categories such as "residential area" group show 500 of variance level with less than 200 pixel size show homogeneous unit of the land cover. About 100 types of land cover objects are classified by the supervise classification process, including unclassified objects
关键词：Object oriented classification; High resolution Satellite Image; Land use
