Band Analysis for Land Use in Multi Spectral Images

Hyper spectral and multi spectral image analysis is the commonly used technique for land use and land cover classification. Effective use of the land cover can play a vital role in the development of country. Multi spectral satellites use passive sensor, hence the only source of energy involved in the acquisition of satellite imagery is the reflectance of the sun. In order to investigate the role of individual bands of the Visible and infra-red region in the recognition of land covers such as vegetation, non-vegetation, settlements and barren land an extensive research has been carried out. This paper is focused in the dissection and contribution of individual component (band) of SPOT-5 imagery for land cover analysis as well. In this article extensive experimentation has been carried out which reveals the effect of individual and combine bands in the recognition of land cover. Classifications of various bands were done using supervised machine learning classification, random forest classifier has been used for classification purpose.