A genetic fuzzy rule-based classifier for land cover image classification

This paper proposes the use of a boosted genetic fuzzy classifier (BGFC) for land cover classification from multispectral images. The model´s learning algorithm is divided into two stages. The first stage iteratively generates fuzzy rules, employing a boosting algorithm that localizes new rules in uncovered subspaces of the feature space. Each rule is obtained through an efficient genetic rule extraction method, which both adapts the parameters of the fuzzy sets in the premise space and determines the required features of the rule, further improving the interpretability of the obtained model. The second stage fine-tunes the obtained rule base through an evolutionary algorithm (EA), improving the cooperation among the fuzzy rules and, thus, increasing the classification performance attained after the first stage. The BGFC is tested using an IKONOS multispectral VHR image, in the agricultural area surrounding a lake-wetland ecosystem in northern Greece. The results indicate that the proposed system is able to handle multi-dimensional feature spaces, effectively exploiting information from different feature sources.