A hierarchical fuzzy clustering algorithm using pyramid linking data and land-cover classification experiments

This paper proposes a hierarchical fuzzy clustering algorithm using pyramid linking data. The authors implemented computational code of the hierarchical fuzzy clustering scheme on a Sun workstation. The code incorporates flexible computational modes. The authors evaluated the code experimentally using hypothetical data and image data from Earth observation satellites. Three levels of a pyramid linking data system from these data are generated using 1 to 16 neighbor pixel averaging according to the pyramid linking information. The hypothetical data have been efficiently segmented. Some test sites of the satellite images are experimentally classified by this code.