How to handle spatial correlations in SAR despeckling? Resampling strategies and deep learning approaches

Emanuele Dalsasso; Loïc Denis; Florence Tupin

How to handle spatial correlations in SAR despeckling? Resampling strategies and deep learning approaches

Emanuele Dalsasso
, Loïc Denis
, Florence Tupin

Institut Polytechnique de Paris
Centre national de la recherche scientifique

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Speckle noise strongly affects Synthetic Aperture Radar (SAR) images, causing strong intensity fluctuations that make them difficult to analyze. Although many speckle reduction algorithms have been proposed, how to effectively deal with the spatial correlations of speckle remains an open question, especially in the most recent deep learning approaches. This paper tries to address this problem. Existing approaches to tackle the speckle correlations are described. Then, a standard training strategy for deep learning is proposed. Two models are trained and the increased robustness brought by including a Total Variation (TV) term in the loss function is analyzed on Sentinel-1 images.

Original language	English
Title of host publication	EUSAR 2021 - 13th European Conference on Synthetic Aperture Radar, Proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	1233-1238
Number of pages	6
ISBN (Electronic)	9783800754571
Publication status	Published - 1 Jan 2021
Event	13th European Conference on Synthetic Aperture Radar, EUSAR 2021 - Virtual, Online, Germany Duration: 29 Mar 2021 → 1 Apr 2021

Publication series

Name	Proceedings of the European Conference on Synthetic Aperture Radar, EUSAR
Volume	2021-March
ISSN (Print)	2197-4403

Conference

Conference	13th European Conference on Synthetic Aperture Radar, EUSAR 2021
Country/Territory	Germany
City	Virtual, Online
Period	29/03/21 → 1/04/21

Cite this

Dalsasso, E., Denis, L., & Tupin, F. (2021). How to handle spatial correlations in SAR despeckling? Resampling strategies and deep learning approaches. In EUSAR 2021 - 13th European Conference on Synthetic Aperture Radar, Proceedings (pp. 1233-1238). (Proceedings of the European Conference on Synthetic Aperture Radar, EUSAR; Vol. 2021-March). Institute of Electrical and Electronics Engineers Inc..

Dalsasso, Emanuele ; Denis, Loïc ; Tupin, Florence. / How to handle spatial correlations in SAR despeckling? Resampling strategies and deep learning approaches. EUSAR 2021 - 13th European Conference on Synthetic Aperture Radar, Proceedings. Institute of Electrical and Electronics Engineers Inc., 2021. pp. 1233-1238 (Proceedings of the European Conference on Synthetic Aperture Radar, EUSAR).

@inproceedings{7f59d3c9c07f4c2f8ae9cbaa3d64a937,

title = "How to handle spatial correlations in SAR despeckling? Resampling strategies and deep learning approaches",

abstract = "Speckle noise strongly affects Synthetic Aperture Radar (SAR) images, causing strong intensity fluctuations that make them difficult to analyze. Although many speckle reduction algorithms have been proposed, how to effectively deal with the spatial correlations of speckle remains an open question, especially in the most recent deep learning approaches. This paper tries to address this problem. Existing approaches to tackle the speckle correlations are described. Then, a standard training strategy for deep learning is proposed. Two models are trained and the increased robustness brought by including a Total Variation (TV) term in the loss function is analyzed on Sentinel-1 images.",

author = "Emanuele Dalsasso and Lo{\"i}c Denis and Florence Tupin",

note = "Publisher Copyright: {\textcopyright} VDE VERLAG GMBH Berlin Offenbach; 13th European Conference on Synthetic Aperture Radar, EUSAR 2021 ; Conference date: 29-03-2021 Through 01-04-2021",

year = "2021",

month = jan,

day = "1",

language = "English",

series = "Proceedings of the European Conference on Synthetic Aperture Radar, EUSAR",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "1233--1238",

booktitle = "EUSAR 2021 - 13th European Conference on Synthetic Aperture Radar, Proceedings",

}

Dalsasso, E, Denis, L & Tupin, F 2021, How to handle spatial correlations in SAR despeckling? Resampling strategies and deep learning approaches. in EUSAR 2021 - 13th European Conference on Synthetic Aperture Radar, Proceedings. Proceedings of the European Conference on Synthetic Aperture Radar, EUSAR, vol. 2021-March, Institute of Electrical and Electronics Engineers Inc., pp. 1233-1238, 13th European Conference on Synthetic Aperture Radar, EUSAR 2021, Virtual, Online, Germany, 29/03/21.

How to handle spatial correlations in SAR despeckling? Resampling strategies and deep learning approaches. / Dalsasso, Emanuele; Denis, Loïc; Tupin, Florence.
EUSAR 2021 - 13th European Conference on Synthetic Aperture Radar, Proceedings. Institute of Electrical and Electronics Engineers Inc., 2021. p. 1233-1238 (Proceedings of the European Conference on Synthetic Aperture Radar, EUSAR; Vol. 2021-March).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - How to handle spatial correlations in SAR despeckling? Resampling strategies and deep learning approaches

AU - Dalsasso, Emanuele

AU - Denis, Loïc

AU - Tupin, Florence

PY - 2021/1/1

Y1 - 2021/1/1

N2 - Speckle noise strongly affects Synthetic Aperture Radar (SAR) images, causing strong intensity fluctuations that make them difficult to analyze. Although many speckle reduction algorithms have been proposed, how to effectively deal with the spatial correlations of speckle remains an open question, especially in the most recent deep learning approaches. This paper tries to address this problem. Existing approaches to tackle the speckle correlations are described. Then, a standard training strategy for deep learning is proposed. Two models are trained and the increased robustness brought by including a Total Variation (TV) term in the loss function is analyzed on Sentinel-1 images.

AB - Speckle noise strongly affects Synthetic Aperture Radar (SAR) images, causing strong intensity fluctuations that make them difficult to analyze. Although many speckle reduction algorithms have been proposed, how to effectively deal with the spatial correlations of speckle remains an open question, especially in the most recent deep learning approaches. This paper tries to address this problem. Existing approaches to tackle the speckle correlations are described. Then, a standard training strategy for deep learning is proposed. Two models are trained and the increased robustness brought by including a Total Variation (TV) term in the loss function is analyzed on Sentinel-1 images.

UR - https://www.scopus.com/pages/publications/85106025522

M3 - Conference contribution

AN - SCOPUS:85106025522

T3 - Proceedings of the European Conference on Synthetic Aperture Radar, EUSAR

SP - 1233

EP - 1238

BT - EUSAR 2021 - 13th European Conference on Synthetic Aperture Radar, Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 13th European Conference on Synthetic Aperture Radar, EUSAR 2021

Y2 - 29 March 2021 through 1 April 2021

ER -

Dalsasso E, Denis L, Tupin F. How to handle spatial correlations in SAR despeckling? Resampling strategies and deep learning approaches. In EUSAR 2021 - 13th European Conference on Synthetic Aperture Radar, Proceedings. Institute of Electrical and Electronics Engineers Inc. 2021. p. 1233-1238. (Proceedings of the European Conference on Synthetic Aperture Radar, EUSAR).

How to handle spatial correlations in SAR despeckling? Resampling strategies and deep learning approaches

Abstract

Publication series

Conference

Other files and links

Fingerprint

Cite this