REDUCING BOUNDARY EFFECTS IN A KERNEL-BASED CLASSIFIER

In this letter, a method is presented to reduce misclassification at t he boundaries between different types of land covers. Errors are cause d by use of spatial features extracted from pixel neighbourhoods. Simp le thresholding and region-growing techniques are used to reduce such errors. The method is included as an extension to a frequency-based co ntextual classifier (FBC). Some experiments have been carried out to e valuate both the original classifier, FBC, and the extended version us ing an image acquired with a Compact Airborne Spectrographic Imager (C ASI). Experimental results show that overall classification accuracies with the CASI image are 84.80 per cent and 89-37 per cent obtained us ing the FBC and the extended FBC, respectively.