A hybrid approach to fuzzy land cover classification

We propose here a fuzzy hybrid methodology for the classification, conceived as a cognitive process, of remote sensing images. The salient aspect of the approach is the combined use of different techniques: the linear mixture model, a supervised fuzzy statistical classifier and a fuzzy labeling technique, An application for the identification of rice crops in a Landsat Thematic Mapper image has been developed with the aim of experimentally evaluating the performance of the overall strategy in a real domain where fuzzy membership to classes are essential in class discrimination, The results have then been compared with those obtained by means of the Maximum Likelihood classifier.