Research Article
SAR Image Despeckling using a Convolutional Neural Network
@INPROCEEDINGS{10.4108/eai.28-4-2025.2357927, author={P Vasundaradevi and Kandasamy Sellamuthu and Manoj Kumar R and Deepak K V and Mathan M}, title={SAR Image Despeckling using a Convolutional Neural Network}, proceedings={Proceedings of the 4th International Conference on Information Technology, Civil Innovation, Science, and Management, ICITSM 2025, 28-29 April 2025, Tiruchengode, Tamil Nadu, India, Part I}, publisher={EAI}, proceedings_a={ICITSM PART I}, year={2025}, month={10}, keywords={synthetic aperture radar (sar) speckle noise reduction convolutional neural networks (cnns) despeckling image enhancement deep learning in remote sensing}, doi={10.4108/eai.28-4-2025.2357927} }- P Vasundaradevi
Kandasamy Sellamuthu
Manoj Kumar R
Deepak K V
Mathan M
Year: 2025
SAR Image Despeckling using a Convolutional Neural Network
ICITSM PART I
EAI
DOI: 10.4108/eai.28-4-2025.2357927
Abstract
This paper aims to present an investigation into how CNNs can do their task to clear the speckle noise from SAR images efficiently. One pair of Sentinel-1 and Sentinel-2 image data was selected for training in preprocessed optical pictures, and image noise was added during augmentation. The CNN architecture employed convolutional layers with included features like batch normalization, ReLU, and LeakyReLU as activation functions for speeding despeckling. Some of the tests will yield results showing how the proposed method improves the visual clarity of SAR images without losing important structural information. It represents an approach that is able to replace the traditional filters on the basis of deep learning for many applications of SAR images interpretation, including agricultural assessment, disaster monitoring, and land cover classification.
