Results and Disscussions
1.3.2. Results of Supervised Classification of LISS III Image
The LISS III image was classified into same eight classes as mentioned above (healthy tea patches, moderately affected patches, diseased tea patches, barren land, river, river bed, scrubs and settlements) as shown in Figure 18. The inclusion of more classes of different vegetation types leads to confusion with other classes and misclassification with classes mixing with each other. Of the three classification techniques MLC showed the lowest degree of mixing. The classified image was validated with the ground truth data. The best result was obtained from maximum likelihood classifier (MLC). The overall accuracy for the LISS III classified image was found to be 86.13%.

Figure 18: Landuse/Landcover Map of Sonitpur District (LISS III, February 2004)