International Journal of Applied Earth Observation and Geoinformation a Study of Supervised Classification Accuracy in Fuzzy Topological Methods

Abstract The multiple classifier system (MCS) is an effective automatic classification method, useful in connection with remote sensing analysis techniques. Combining MSC with induced fuzzy topology enables a decomposition of image classes. This fuzzy topological MCS then provides a new and improved approach to classification. The basic classification methods discussed in this paper include maximum likelihood classification (MLC), minimum distance classification (MIND) and Mahalanobis distance classification (MAH). In this paper, the use of the fuzzy topology techniques in combination with the current classification methods is discussed. The methods included are (1) ordinary single classifier classification methods; (2) fuzzy single classifier classification methods; (3) simple average MCS; (4) fuzzy topological simple average MCS; (5) eigen-value MCS; (6) fuzzy topology and eigen-values MCS. This new experimental approach, involving such combinations for comparing the kappa values and overall accuracies is also discussed. After comparing the kappa values and overall accuracies of these classification methods, the experimental results, demonstrated that (a) methods combining with fuzzy topology concepts produced better classification accuracy than the ordinary methods; (b) the eigen-value MCS method produces better classification accuracy than the non-fuzzy method and (c) the best classifier combination was found to be MLC + MIND + MAH fuzzy eigen-value MCS.