Water body area identification and classification using hyperspectral images

The water resource monitoring system based on satellite images has a significant drawback, that it does not deliver reliable data. Main drawbacks in the existing methodologies are, the water resources and their existence could not be correctly recovered, and the results obtained were not conspicuous. The proposed system properly retrieves the existence of water resources and allows monitoring using satellite images. In the proposed system, deep learning based segmentation and SVM classifier have been adopted, which properly retrieves the existence of water resources. The proposed method has been applied on the pond in Karur district and has been found that, 8% of water resource has been reduced during 2019-2020. The proposed method produces the outcomes with the Accuracy of 98% when compared with other existing methods. This method may be used to track water resources with ranges in the future, as well as monitor occupied spots in water resources throughout all districts.