Potential of Hyperspectral and Multi-angle Chris/proba Images in Vegetation Identification and Monitoring

This contribution reports on the results obtained in mapping vegetated surfaces within the target site of Fascati/Tor Vergata, by the use of the CHRIS Proba images. In particular our purpose is to show how the use of multi-angular observations, properly elaborated and rectified, can provide further pieces of information for the vegetation identification with respect to those available considering only one acquisition angle, and how this information enhancement improves the final classification product. To this latter purpose, a pixelbased neural network approach has been implemented. Its assessment has been made by means of visual inspection and quantified by the use of the confusion matrix. A brief introduction to the image correction is also reported and explained.