Forest classification using spectrometer and SAR data

This work deals with automatic classification of forest areas using remote sensing imagery. The authors compare the discrimination ability of two complementary sensors, a SAR sensor and a spectrometer. A Gaussian maximum likelihood classifier was used in all classification experiments. The hyperspectral data alone gave fairly good results for classification of tree species. The results for SAR data alone were not convincing. Joining the two data set in a simple fusion experiment improved the results obtained significantly for data from a single sensor, and also allowed a classification of tree species and height simultaneously