Despeckling Sentinel-1 GRD Images by Deep-Learning and Application to Narrow River Segmentation

This paper presents a despeckling method for Sentinel-1 GRD images based on the recently proposed framework "SAR2SAR": a self-supervised training strategy. Training the deep neural network on collections of Sentinel 1 GRD images leads to a despeckling algorithm that is robust to space-vari...

Full description

Saved in:
Bibliographic Details
Main Authors: Gasnier, Nicolas, Dalsasso, Emanuele, Denis, Loic, Tupin, Florence
Format: Conference Proceeding
Language:English
Subjects:
Correlation
deep learning
denoising
Detectors
Image segmentation
Noise reduction
SAR
Sentinel-1
Speckle
Time series analysis
Training
water detection
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper presents a despeckling method for Sentinel-1 GRD images based on the recently proposed framework "SAR2SAR": a self-supervised training strategy. Training the deep neural network on collections of Sentinel 1 GRD images leads to a despeckling algorithm that is robust to space-variant spatial correlations of speckle. Despeckled images improve the detection of structures like narrow rivers. We apply a detector based on exogenous information and a linear features detector and show that rivers are better segmented when the processing chain is applied to images pre-processed by our despeckling neural network.
ISSN:2153-7003
DOI:10.1109/IGARSS47720.2021.9554350