Decision fusion, classification of multisource data

Abstract

In this entry on decision fusion for classification of multisource remote sensing data, a brief definition on data fusion and an overview of multisource remote sensing applications was given. Examples of how approaches, which are based on decision fusion, can be applied to currently typical remote sensing data sets were presented. In general, these studies show that the accuracy of remote sensing land cover classifications is increased due to the rapid development of EO systems, resulting in the availability of multitemporal, multisource, or high-dimensional data sets on the one hand and the use of adequate classification approaches on the other. Recent approaches, for example, based on decision fusion, enable the handling of such complex data sets efficiently and in a robust manner. Problems that occur due to limitations of traditional methods can be avoided, and results are consequently more accurate. A higher number of diverse EO instruments and further enhancement of these systems is expected for the future. Therefore, appropriate data fusion and classification of multisource data is an important ongoing research topic in the field of remote sensing. The methodological development outlined in this entry is an alternative direction to continue.