||Hyperspectral data are remotely sensed image sets obtained by measuring the light reflected from a scene within hundreds of narrow contiguous spectral wavelength intervals. Due to its richness in information, hyperspectral imagery allows for detection of targets covering areas smaller than a pixel or separation of objects and shapes otherwise undistinguishable in regular images.
The talk will provide an overview of hyperspectral images, current sensor technology and applications. Next it will discuss new results focusing on improving efficiency in processing by using high performance computing. Most of the hyperspectral image processing techniques have complexity that depends directly on the number of spectral bands in the acquired data. Since this is usually a large number, it is of interest to find methods that transform the data cube into one with reduced dimensionality while, at the same time, maintaining as much information content as possible. Finally we will discuss the use of hyperspectral imaging for face recognition using both spatial and spectral information.