Land use pattern classification of remote sensing imagery data is imperative to research that is used in remote sensing applications. Remote sensing (RS) technologies were exploited to mine some of the significant spatially variable factors, such as land cover and land use (LCLU), from satellite images of remote arid areas in Karnataka, India. Four diverse classification techniques unsupervised, and supervised (Maximum likelihood, Mahalnobis Distance, and Minimum Distance) are applied in Bellary district in Karnataka, India for the classification of the raw satellite images. The developed maps are then visually compared with each other and accuracy evaluations make using of ground-truths are carried out. It was initiated that the Maximum likelihood technique gave the finest results and both Minimum distance and Mahalnobis distance methods overvalued agricultural land areas. In spite of missing a few insignificant features due to the low resolution of the satellite images, a high-quality accord between parameters extracted automatically from the developed maps and field observations was found. Keywords: Remote sensing (RS), land cover and land use (LCLU).
Read full abstract