A Wavelet Transform Applied Spectral Index for Effective Water Body Extraction from Moderate-Resolution Satellite Images

Abstract

The outbreak of pollution due to rapid urbanization has led to a huge threat unto the ecosystem where the water reserves are severely affected. In order to protect these coastal regions, drinking water resources, and artificial reservoirs, the satellite image-based monitoring models could be deployed. For developing such models using moderate-resolution data, the extraction of water body from the large land cover would be more complex. This limitation of effective water body extraction has been addressed in this paper through using the combination of wavelets and image transformation methods. A novel Wavelet-based Water Index (WaWI) has been proposed, and the results achieved are quantitatively assessed with both the spectral- and clustering-based water body extraction results such as Normalized Difference Moisture Index (NDMI), Modified Normalized Difference Moisture Index (MNDMI), and K-means clustering-based water body extraction using the Image Quality Assessment metrics like correlation coefficient, structural similarity index (SSIM), and Jaccard’s similarity measures. The results achieved profoundly justifies the effectiveness of WaWI in water body extraction.