Information and understanding: Analysis of remotely sensed data

A review is given of the development of the field of image understanding in remote sensing, with an emphasis on the contributions of David Landgrebe and his group at the Laboratory for Applications of Remote Sensing, Purdue University. The differences in approach required for multispectral, hyperspectral and radar image data are emphasised, in which the seminal contributions to the field by Landgrebe as well as others are summarised. The treatment concludes by examining the current problem of thematic mapping from mixed spatial data types, such as would be found in a geographical information system. Rather than seeking techniques that "fuse" available data types as a means for deriving joint inferences, it is proposed instead that the most practical means is to have each individual data source analysed separately by the most appropriate techniques and the fuse at the label level using the facilities of an expert consultant.