Satellite Image Classification for Detecting Unused Landscape using CNN

As the landscapes changes day by day it leads to the increasing use of unused lands, by which unused lands can be used for various purposes like agriculture, developing city infrastructure and many more. This paper helps in automating the process of detecting the unused land space. In this work, a system for satellite image processing that detects unused land is proposed. Here remote sensing earth images are taken as the dataset where the pre-processing step includes converting image into greyscale image, compression and noise removal. Segmentation is done to partition the region of used and unused lands. Feature extraction is done here using local binary feature extraction in-order to identify edge, flat and corner surfaces. As the mentioned various algorithm is used in classification and labeling of remote sensing earth images. CNN algorithm is also used for classification and labeling of classification is done automatically by the use of CNN algorithm. Random forest is used to segregate two landscapes as used and unused land which gives accuracy better than the existing systems.