HAZE MITIGATION IN HIGH-RESOLUTION SATELLITE IMAGERY USING ENHANCED STYLE-TRANSFER NEURAL NETWORK AND NORMALIZATION ACROSS MULTIPLE GPUS

Byung H. Park; Somrita Chattopadhyay; John Burgin

doi:10.1109/IGARSS47720.2021.9553574

HAZE MITIGATION IN HIGH-RESOLUTION SATELLITE IMAGERY USING ENHANCED STYLE-TRANSFER NEURAL NETWORK AND NORMALIZATION ACROSS MULTIPLE GPUS

Byung H. Park, Somrita Chattopadhyay, John Burgin

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

1 Scopus citations

Abstract

Despite recent advances in deep learning approaches, haze mitigation in large satellite images is still a challenging problem. Due to amorphous nature of haze, object detection or image segmentation approaches are not applicable. Also it is practically infeasible to obtain ground truths for training. Bounded memory capacity of GPUs is another constraint that limits the size of image to be processed. In this paper, we propose a style transfer based neural network approach to mitigate haze in a large overhead imagery. The network is trained without paired ground truths; further, perception loss is added to restore vivid colors, enhance contrast and minimize artifacts. The paper also illustrates our use of multiple GPUs in a collective way to produce a single coherent clear image where each GPU dehazes different portions of a large hazy image.

Original language	English
Title of host publication	IGARSS 2021 - 2021 IEEE International Geoscience and Remote Sensing Symposium, Proceedings
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	2827-2830
Number of pages	4
ISBN (Electronic)	9781665403696
DOIs	https://doi.org/10.1109/IGARSS47720.2021.9553574
State	Published - 2021
Event	2021 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2021 - Brussels, Belgium Duration: Jul 12 2021 → Jul 16 2021

Publication series

Name	International Geoscience and Remote Sensing Symposium (IGARSS)
Volume	2021-July

Conference

Conference	2021 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2021
Country/Territory	Belgium
City	Brussels
Period	07/12/21 → 07/16/21

Funding

*This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript or allow others to do so, for the US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).

Keywords

Convolutional neural networks
CycleGAN
Haze mitigation
Remote sensing

Access to Document

10.1109/IGARSS47720.2021.9553574

Cite this

Park, B. H., Chattopadhyay, S., & Burgin, J. (2021). HAZE MITIGATION IN HIGH-RESOLUTION SATELLITE IMAGERY USING ENHANCED STYLE-TRANSFER NEURAL NETWORK AND NORMALIZATION ACROSS MULTIPLE GPUS. In IGARSS 2021 - 2021 IEEE International Geoscience and Remote Sensing Symposium, Proceedings (pp. 2827-2830). (International Geoscience and Remote Sensing Symposium (IGARSS); Vol. 2021-July). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IGARSS47720.2021.9553574

Park, Byung H. ; Chattopadhyay, Somrita ; Burgin, John. / HAZE MITIGATION IN HIGH-RESOLUTION SATELLITE IMAGERY USING ENHANCED STYLE-TRANSFER NEURAL NETWORK AND NORMALIZATION ACROSS MULTIPLE GPUS. IGARSS 2021 - 2021 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. Institute of Electrical and Electronics Engineers Inc., 2021. pp. 2827-2830 (International Geoscience and Remote Sensing Symposium (IGARSS)).

@inproceedings{03fd8fd15b1c460597e77f16beebfb57,

title = "HAZE MITIGATION IN HIGH-RESOLUTION SATELLITE IMAGERY USING ENHANCED STYLE-TRANSFER NEURAL NETWORK AND NORMALIZATION ACROSS MULTIPLE GPUS",

abstract = "Despite recent advances in deep learning approaches, haze mitigation in large satellite images is still a challenging problem. Due to amorphous nature of haze, object detection or image segmentation approaches are not applicable. Also it is practically infeasible to obtain ground truths for training. Bounded memory capacity of GPUs is another constraint that limits the size of image to be processed. In this paper, we propose a style transfer based neural network approach to mitigate haze in a large overhead imagery. The network is trained without paired ground truths; further, perception loss is added to restore vivid colors, enhance contrast and minimize artifacts. The paper also illustrates our use of multiple GPUs in a collective way to produce a single coherent clear image where each GPU dehazes different portions of a large hazy image.",

keywords = "Convolutional neural networks, CycleGAN, Haze mitigation, Remote sensing",

author = "Park, {Byung H.} and Somrita Chattopadhyay and John Burgin",

note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; 2021 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2021 ; Conference date: 12-07-2021 Through 16-07-2021",

year = "2021",

doi = "10.1109/IGARSS47720.2021.9553574",

language = "English",

series = "International Geoscience and Remote Sensing Symposium (IGARSS)",

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

pages = "2827--2830",

booktitle = "IGARSS 2021 - 2021 IEEE International Geoscience and Remote Sensing Symposium, Proceedings",

}

Park, BH, Chattopadhyay, S & Burgin, J 2021, HAZE MITIGATION IN HIGH-RESOLUTION SATELLITE IMAGERY USING ENHANCED STYLE-TRANSFER NEURAL NETWORK AND NORMALIZATION ACROSS MULTIPLE GPUS. in IGARSS 2021 - 2021 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. International Geoscience and Remote Sensing Symposium (IGARSS), vol. 2021-July, Institute of Electrical and Electronics Engineers Inc., pp. 2827-2830, 2021 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2021, Brussels, Belgium, 07/12/21. https://doi.org/10.1109/IGARSS47720.2021.9553574

HAZE MITIGATION IN HIGH-RESOLUTION SATELLITE IMAGERY USING ENHANCED STYLE-TRANSFER NEURAL NETWORK AND NORMALIZATION ACROSS MULTIPLE GPUS. / Park, Byung H.; Chattopadhyay, Somrita; Burgin, John.
IGARSS 2021 - 2021 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. Institute of Electrical and Electronics Engineers Inc., 2021. p. 2827-2830 (International Geoscience and Remote Sensing Symposium (IGARSS); Vol. 2021-July).

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

TY - GEN

T1 - HAZE MITIGATION IN HIGH-RESOLUTION SATELLITE IMAGERY USING ENHANCED STYLE-TRANSFER NEURAL NETWORK AND NORMALIZATION ACROSS MULTIPLE GPUS

AU - Park, Byung H.

AU - Chattopadhyay, Somrita

AU - Burgin, John

PY - 2021

Y1 - 2021

N2 - Despite recent advances in deep learning approaches, haze mitigation in large satellite images is still a challenging problem. Due to amorphous nature of haze, object detection or image segmentation approaches are not applicable. Also it is practically infeasible to obtain ground truths for training. Bounded memory capacity of GPUs is another constraint that limits the size of image to be processed. In this paper, we propose a style transfer based neural network approach to mitigate haze in a large overhead imagery. The network is trained without paired ground truths; further, perception loss is added to restore vivid colors, enhance contrast and minimize artifacts. The paper also illustrates our use of multiple GPUs in a collective way to produce a single coherent clear image where each GPU dehazes different portions of a large hazy image.

AB - Despite recent advances in deep learning approaches, haze mitigation in large satellite images is still a challenging problem. Due to amorphous nature of haze, object detection or image segmentation approaches are not applicable. Also it is practically infeasible to obtain ground truths for training. Bounded memory capacity of GPUs is another constraint that limits the size of image to be processed. In this paper, we propose a style transfer based neural network approach to mitigate haze in a large overhead imagery. The network is trained without paired ground truths; further, perception loss is added to restore vivid colors, enhance contrast and minimize artifacts. The paper also illustrates our use of multiple GPUs in a collective way to produce a single coherent clear image where each GPU dehazes different portions of a large hazy image.

KW - Convolutional neural networks

KW - CycleGAN

KW - Haze mitigation

KW - Remote sensing

UR - http://www.scopus.com/inward/record.url?scp=85126025657&partnerID=8YFLogxK

U2 - 10.1109/IGARSS47720.2021.9553574

DO - 10.1109/IGARSS47720.2021.9553574

M3 - Conference contribution

AN - SCOPUS:85126025657

T3 - International Geoscience and Remote Sensing Symposium (IGARSS)

SP - 2827

EP - 2830

BT - IGARSS 2021 - 2021 IEEE International Geoscience and Remote Sensing Symposium, Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2021 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2021

Y2 - 12 July 2021 through 16 July 2021

ER -

Park BH, Chattopadhyay S, Burgin J. HAZE MITIGATION IN HIGH-RESOLUTION SATELLITE IMAGERY USING ENHANCED STYLE-TRANSFER NEURAL NETWORK AND NORMALIZATION ACROSS MULTIPLE GPUS. In IGARSS 2021 - 2021 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. Institute of Electrical and Electronics Engineers Inc. 2021. p. 2827-2830. (International Geoscience and Remote Sensing Symposium (IGARSS)). doi: 10.1109/IGARSS47720.2021.9553574

HAZE MITIGATION IN HIGH-RESOLUTION SATELLITE IMAGERY USING ENHANCED STYLE-TRANSFER NEURAL NETWORK AND NORMALIZATION ACROSS MULTIPLE GPUS

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this