Categorization of the pixels in a digital image representing the ground features into some user defined classes is the process of image classification. Two common methods are available for image classification, viz., (i) manual (or on-screen) digitization and (ii) digital classification. Digital classification can be further grouped into supervised and unsupervised classification. A suitable classification scheme is a pre-requisite for meaningful representation of existing ground features in digital format. Consideration of classification scheme depends on factors like user's needs, characteristics of the study area and nature of remote sensing data [24]. Supervised digital classification method is found suitable for image classification in the present study. Based on the need and objectives, two classes viz. (i) crop area, and (ii) non-crop area are considered
for classification. The identification of crop area enables
assessment of potential power generation from surplus rice
straw. The spatial distribution of non-crop area needs further
analysis in respect to utilization of decentralized power. The
steps followed to classify the image are briefly described
below.