The pre-processed images are then classified by both unsupervised,
supervised classification methods. In un-supervised
classification method the ISODATA clustering algorithm
which is built in the ERDAS Imagine will classify
according to the number of classes required and the digital
number of the pixels available. In the supervised classification
technique the maximum likely hood algorithm will
classify the image based on the training sets (signatures)
provided by the user based on his field knowledge. The
training data given by the user guides the software as to
what types of pixels are to be selected for certain land cover
type. The un-supervised classified image has been used for
reference and for understanding about the distribution of
pixels with different digital numbers. The classification
finally gives the land use/land cover image of the area.
Four land cover classes namely agricultural land, built up
area, barren land and water bodies are identified in the
study area.