A1 Refereed original research article in a scientific journal
ECDSA-Based Water Bodies Prediction from Satellite Images with UNet
Authors: Ch Anusha, Ch Rupa, Gadamsetty Samhitha, Iwendi Celestine, Gadekallu Thippa Reddy, Ben Dhaou Imed
Publisher: MDPI
Publication year: 2022
Journal: Water
Journal name in source: WATER
Journal acronym: WATER-SUI
Article number: 2234
Volume: 14
Issue: 14
Number of pages: 22
DOI: https://doi.org/10.3390/w14142234
Web address : https://www.mdpi.com/2073-4441/14/14/2234
Self-archived copy’s web address: https://research.utu.fi/converis/portal/detail/Publication/176159933
The detection of water bodies from satellite images plays a vital role in research development. It has a wide range of applications such as the prediction of natural disasters and detecting drought and flood conditions. There were few existing applications that focused on detecting water bodies that are becoming extinct in nature. The dataset to train this deep learning model is taken from Kaggle. It has two classes, namely water bodies and masks. There is a total of 2841 sentinel-2 satellite images with corresponding 2841 masks. Additionally, the present work focuses on using UNet, Tensorflow to detect the water bodies. It uses a Nadam optimizer to reduce the losses. It also finds best-optimized parameters for the activation function, a number of nodes in each layer. This proposed model achieves integrity by embedding a security feature Elliptic Curve Digital Signature Algorithm (ECDSA). It generates a digital signature for the predicted area of water bodies which helps to secure the key and the detected water bodies while transmitting in a channel. Thus, the proposed model ensures the performance accuracy of 94% which can also work the same for edge detection, detection in blurred and low-resolution images. The model is highly robust.
Downloadable publication This is an electronic reprint of the original article. |  |
