Image classification using visible RGB bands

The purpose of this method is to classify the image pixels in to various land cover categories or classes or themes. These themes can later be studied individually. Usually multispectral data is used to understand and classify an image. The problem is that images generally comprises of only RGB bands. NIR band is not always available. So the objective of this proposed algorithm is to perform image classification using visible Red, Green and blue bands without converting RGB image into any other color model. This paper presents classification formula for land covers such as vegetation, water, concrete road and brown build-up area.