SAR TO OPTICAL IMAGE SYNTHESIS FOR CLOUD REMOVAL WITH GENERATIVE ADVERSARIAL NETWORKS
Optical imagery is often affected by the presence of clouds. Aiming to reduce their effects, different reconstruction techniques have been proposed in the last years. A common alternative is to extract data from active sensors, like Synthetic Aperture Radar (SAR), because they are almost independent...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Copernicus Publications
2018-09-01
|
Series: | ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
Online Access: | https://www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net/IV-1/5/2018/isprs-annals-IV-1-5-2018.pdf |
id |
doaj-d323fc3dc79c45218d26ea848ff609a3 |
---|---|
record_format |
Article |
spelling |
doaj-d323fc3dc79c45218d26ea848ff609a32020-11-25T00:56:19ZengCopernicus PublicationsISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences2194-90422194-90502018-09-01IV-151110.5194/isprs-annals-IV-1-5-2018SAR TO OPTICAL IMAGE SYNTHESIS FOR CLOUD REMOVAL WITH GENERATIVE ADVERSARIAL NETWORKSJ. D. Bermudez0P. N. Happ1D. A. B. Oliveira2R. Q. Feitosa3R. Q. Feitosa4Pontifical Catholic University of Rio de Janeiro, BrazilPontifical Catholic University of Rio de Janeiro, BrazilIBM ResearchPontifical Catholic University of Rio de Janeiro, BrazilRio de Janeiro State University, BrazilOptical imagery is often affected by the presence of clouds. Aiming to reduce their effects, different reconstruction techniques have been proposed in the last years. A common alternative is to extract data from active sensors, like Synthetic Aperture Radar (SAR), because they are almost independent on the atmospheric conditions and solar illumination. On the other hand, SAR images are more complex to interpret than optical images requiring particular handling. Recently, Conditional Generative Adversarial Networks (cGANs) have been widely used in different image generation tasks presenting state-of-the-art results. One application of cGANs is learning a nonlinear mapping function from two images of different domains. In this work, we combine the fact that SAR images are hardly affected by clouds with the ability of cGANS for image translation in order to map optical images from SAR ones so as to recover regions that are covered by clouds. Experimental results indicate that the proposed solution achieves better classification accuracy than SAR based classification.https://www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net/IV-1/5/2018/isprs-annals-IV-1-5-2018.pdf |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
J. D. Bermudez P. N. Happ D. A. B. Oliveira R. Q. Feitosa R. Q. Feitosa |
spellingShingle |
J. D. Bermudez P. N. Happ D. A. B. Oliveira R. Q. Feitosa R. Q. Feitosa SAR TO OPTICAL IMAGE SYNTHESIS FOR CLOUD REMOVAL WITH GENERATIVE ADVERSARIAL NETWORKS ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
author_facet |
J. D. Bermudez P. N. Happ D. A. B. Oliveira R. Q. Feitosa R. Q. Feitosa |
author_sort |
J. D. Bermudez |
title |
SAR TO OPTICAL IMAGE SYNTHESIS FOR CLOUD REMOVAL WITH GENERATIVE ADVERSARIAL NETWORKS |
title_short |
SAR TO OPTICAL IMAGE SYNTHESIS FOR CLOUD REMOVAL WITH GENERATIVE ADVERSARIAL NETWORKS |
title_full |
SAR TO OPTICAL IMAGE SYNTHESIS FOR CLOUD REMOVAL WITH GENERATIVE ADVERSARIAL NETWORKS |
title_fullStr |
SAR TO OPTICAL IMAGE SYNTHESIS FOR CLOUD REMOVAL WITH GENERATIVE ADVERSARIAL NETWORKS |
title_full_unstemmed |
SAR TO OPTICAL IMAGE SYNTHESIS FOR CLOUD REMOVAL WITH GENERATIVE ADVERSARIAL NETWORKS |
title_sort |
sar to optical image synthesis for cloud removal with generative adversarial networks |
publisher |
Copernicus Publications |
series |
ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
issn |
2194-9042 2194-9050 |
publishDate |
2018-09-01 |
description |
Optical imagery is often affected by the presence of clouds. Aiming to reduce their effects, different reconstruction techniques have been proposed in the last years. A common alternative is to extract data from active sensors, like Synthetic Aperture Radar (SAR), because they are almost independent on the atmospheric conditions and solar illumination. On the other hand, SAR images are more complex to interpret than optical images requiring particular handling. Recently, Conditional Generative Adversarial Networks (cGANs) have been widely used in different image generation tasks presenting state-of-the-art results. One application of cGANs is learning a nonlinear mapping function from two images of different domains. In this work, we combine the fact that SAR images are hardly affected by clouds with the ability of cGANS for image translation in order to map optical images from SAR ones so as to recover regions that are covered by clouds. Experimental results indicate that the proposed solution achieves better classification accuracy than SAR based classification. |
url |
https://www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net/IV-1/5/2018/isprs-annals-IV-1-5-2018.pdf |
work_keys_str_mv |
AT jdbermudez sartoopticalimagesynthesisforcloudremovalwithgenerativeadversarialnetworks AT pnhapp sartoopticalimagesynthesisforcloudremovalwithgenerativeadversarialnetworks AT daboliveira sartoopticalimagesynthesisforcloudremovalwithgenerativeadversarialnetworks AT rqfeitosa sartoopticalimagesynthesisforcloudremovalwithgenerativeadversarialnetworks AT rqfeitosa sartoopticalimagesynthesisforcloudremovalwithgenerativeadversarialnetworks |
_version_ |
1725227705727713280 |