Classification of land cover image with spectral signature in machine learning

The land use and land cover maps available from various satellite images of past few decades provide the various landform features. The images are interpreted by their respective spectral reflectance. Individual geomorphic features, the river, lake, hills, forest, soil reflect electromagnetic ray differently with different wavelengths. The spectra appear differently for different objects of landforms. Group of pixels are uniform with respect to the pixel values in several spectral bands. The characteristics of the electromagnetic radiation, its transmission, absorption and reflection on the earth’s surface are used for detection of the objects and classification. Machine learning is an efficient tool to manage the huge volumes of satellite data collected in the form of raster image. The processes of classification of images are done automatically by the classification algorithm by grouping the spectral classes and assigning to the respective landforms on the basis of reflectance property. These form an information class. The aim of the present work is to generate a classified graph on the basis of the spectral signature applied by the classification process of Machine learning. The resultant output is known as spectral response curve. This will help to produce a classified information document of the landform of the land cover raster data.