Segmentation and analysis of hyperspectral data

Summary form only given. We review a previously presented algorithm that segments hyperspectral images on the basis of the two- or three-dimensional histograms of their principal components. Some modifications to improve our previous approach are detailed. After exploring the application of morphology directly to the segmented (digital) images, we focus on the processing of our segmented images in tandem with the original hyperspectral data which produces an "anomaly gray-scale image". Such images, when subject to morphological filtering, prove to be powerful anomaly/target cueing algorithms.