Desenvolupament de tècniques de processament d'imatges hiperespectrals per a teledetecció: codificació i compressió, reconeixement de patrons, transmissió i cooperació a Internet

The main goal of this project is to obtain a compression and classification system for hyperspectral images for its application to the surveillance of several natural resources (Remote Sensing). Hyperspectral images are those obtained by means of a sensor system, either from a satellite or from a plane, that capture the emitted or reflected radiation for different wavelengths or bands. Depending on the number of bands, they are named multi or hyperspectral images. In the present project several computational and statistical aspects related to hyperspectral images are studied: compression, coding, supervised and unsupervised classification, unsupervised segmentation, dimensionality reduction, and feature extraction. Besides, as a technological product, it will be considered to implement several algorithms in a library which could be included in commercial software, in particular, into the ENVI platform. This software library may later be converted into a remote sensing product to be commercialised in the mid-term.