Classification of remotely sensed images using mathematical morphology
Demin Wang, Dong-Chen He, Denis Morin · 2002
Presents the supervised classification result of a remotely sensed image using only its spatial characteristics. A SPOT panchromatic image is decomposed into three scale bands, instead of spectral bands, by multiscale morphological decomposition. Each scale band contains the objects and texture primitives of certain sizes of the image. The classification is performed by applying the maximum-likelihood classifier to local gray level means of the scale bands.>