Cromp and Campbell’s work [7], which examined several image classification algorithms
and argued the need for augmenting the meta-database with information on image content, stands
as the original effort to extend data mining techniques into remote sensing. Alber et al [8]
demonstrated the capability of image search algorithms to search large databases for multi- and
hyperspectral image cubes most closely matching a particular query cube. An interactive search
and analysis tool was presented and tested based on a relevance feedback approach to enhance a
content-based image retrieval process. Chang [9] and Ren [10] studied target detection and image
classification algorithms in hyperspectral imagery. An automatic target generation process
generates a set of targets from image data in an unsupervised manner which is classified by the
target classification process. Schweizer and Moura [11] exploits both the spatial and spectral
correlations in hyperspectral imagery and proposed an approach based on a Gauss–Markov
random field (GMRF) modeling of the clutter, which has the advantage of providing a direct
parameterization of the inverse of the clutter covariance