Key points are not available for this paper at this time.
Our work presents a technique for Detection of Water and Land Surface using Unsupervised Classification on Sentinel C Band Dual Polarimetric SAR Satellite Image in IW mode, at azimuth resolution of 20m and range resolution of 5m. SAR image is pre-processed for Radiometric Corrections, Topographic Corrections and Speckle Filtering. Grey Level Co-occurrence Matrix (GLCM) method extracts Texture Features from SAR image using dual bands, VV and VH. SAR image is classified using GLCM Matrix based on Angular Second Moment (ASM), Mean and Correlation using K Means Clustering to give reasonable class separation for differentiating Land and Water features. Land/Water Mask of classified SAR image is generated and Land/Water Mask of input SAR image is also generated using Shuttle Radar Topography Mission (SRTM) elevation model. Land/Water Mask generated from SRTM data elevation model has been used as Ground Truth for validation. Quantitative analysis of the classified result was done by comparing Land/Water Mask of classified image and Land/Water Mask of input image. Accuracy of classified pixel ranges from 92% to 99% for different regions. The Novelty of our work lies in the GLCM Feature Selection (ASM, Mean, and Correlation) for Land/Water Classification and Validation Procedure followed. The paper also validates the classification results qualitatively by overlaying the classified results on Google map of the same scene. The work has been done using European Space Agency (ESA), Sentinel-1 toolbox (S1TBX) of Sentinel Application Platform (SNAP).Dataset has been acquired from Copernicus Open Archive.
Rani et al. (Thu,) studied this question.