Test of Statistical Methods for the Classification of Polar Terrains on Mars Observed by OMEGA
Résumé
The usual choice of PCA space to analyse remote sensing data is relevant for southern polar cap observations by OMEGA because this method separates signal from noise, especially due to "hot spectels", and reduces data by a factor 255 of 10. However it is not efficient to discriminate albedo/illumination effects from absorption bands. Nevertheless this approach can be improved by a WT filtering.