STATISTICAL, CONNECTIONIST, AND FUZZY INFERENCE TECHNIQUES FOR IMAGE CLASSIFICATION

A spectral classification comparison was performed using four differen t classifiers, the parametric maximum likelihood classifier and three nonparametric classifiers: neural networks, fuzzy rules, and fuzzy neu ral networks. The input image data is a System Pour l'Observation de l a Terre (SPOT) satellite image of Otago Harbour near Dunedin, New Zeal and. The SPOT image data contains three spectral bands in the green, r ed, and visible infrared portions of the electromagnetic spectrum. The specific area contains intertidal vegetation species above and below the waterline. Of specific interest is eelgrass (Zostera novazelandica ), which is a biotic indicator of environmental health. The mixed cove rtypes observed in an in situ field survey are difficult to classify b ecause of subjectivity and water's preferential absorption of the visi ble infrared spectrum. In this analysis, each of the classifiers were applied to the data in two different testing procedures. In the first test procedure, the reference data was divided into training and test by area. Although this is an efficient data handling technique, the cl assifier is not presented with all of the subtle microclimate variatio ns. In the second test procedure, the same reference areas were amalga mated and randomly sorted into training and test data. The amalgamatio n and sorting were performed external to the analysis software. For th e first testing procedure, the highest testing accuracy was obtained t hrough the use of fuzzy inferences at 89%. In the second testing proce dure, the maximum likelihood classifier and the fuzzy neu,al networks provided the best results. Although the testing accuracy for the maxim um likelihood classifier and the fuzzy neural networks were similar, t he latter algorithm has additional features, such as rules extraction, explanation, and fine tuning of individual classes. (C) 1997 SPIE and IS&T.