On the FCC generated, ASTER bands, the tea patches are more prominent than on the LANDSAT and LISS III images. The image was classified using MLC, Minimum Distance to Mean and Parallelepiped classification algorithm into healthy tea patches, moderately healthy tea patches, affected tea patches, settlements, river, river bed, scrubs and barren land as shown in Figure 19. Mixing of classes was comparatively less, in Maximum Likelihood Classifier (MLC) thereby giving the best result. Overall classification accuracy obtained was 87.39%.  Figure 19: Landuse/Landcover Map of Sonitpur District (ASTER, June 2004) From all the three classification techniques it was found that ASTER image with 15m resolution gives good classification results as there is less mixing and less misclassification. |