Modelling uncertainty in high resolution remotely sensed scenes using a fuzzy logic approach

This contribution concentrates on the determination of thematic uncertainty after the classification process. It is shown that in this context –particularly when evaluating remotely sensed scenes showing high spatial resolution – severe problems arise due to indeterminate boundaries in both reference data and classification results. This effect mainly occurs between natural objects or are due to blurred or overlapping definitions of classes or related attributes in a given classification scheme. Based on some approaches in the literature and unsatisfactory tools in existing software packages, we propose the introduction of a new characteristic value, the Fuzzy Certainty Measure (FCM), that is able to model and quantify uncertainty as indeterminate boundaries in reference data and classification results.